Journal of the Mississippi Academy of Sciences Volume 50 July 2005 Number 3

Contents

Journal of the Mississippi Academy of Sciences General Article

Editorial Board 165 The Anthropic Principle, String Theory, and Multiple Universes in Light of the Scientific Method—Jerome Gregorio Begonia Goddard Jackson State University

Maria F.T. Begonia Research Articles Jackson State University

Ibrahim O. Farah 168 Red-cockaded Woodpecker (Picoides borealis) Behavior Jackson State University in a Mississippi Loblolly-Shortleaf Pine Forest—Douglas R. Wood, L. Wesley Burger, Jr., and Francisco J. Vilella Robin Rockhold Univ. of Miss. Medical Center 177 Consumer Preference of Apples Grown in Northern Mississippi—Maria J. Sindoni, Frank B. Matta, and Juan L. Abstracts Editor John Boyle Silva Mississippi State University 183 Modeling Extinction: Density-Dependent Changes in the The Journal of the Mississippi Variance of Population Growth Rates—Thomas E. Academy of Sciences (ISSN 0076- Heering Jr. and David H. Reed 9436) is published in January (annual meeting abstracts), April, July, and October, by the Mississippi Academy Brief Communication of Sciences. Members of the Acad- emy receive the journal as part of 195 New Records for the Phlebotomine Sand their regular (nonstudent) member- ship. Inquiries regarding subscrip- shannoni (Dyar) (Diptera: ) in Missis- tions, availability of back issues, and sippi—Jerome Goddard and Chad P. McHugh address changes should be addressed to The Mississippi Academy of Sci- ences, Post Office Box 55709, Jack- Departments son, MS 39296-5709, telephone 601- 977-0627, or email : msacad@bellsouth. net. 197 President’s Column—Sarah Lea McGuire OFFICERS OF THE MISSISSIPPI ACADEMY OF SCIENCES

President...... Sarah Lea McGuire President-Elect ...... Larry McDaniel Immediate Past-President...... Hamed Benghuzzi Executive Officer...... Charles Swann Junior Academy Director...... Aimée T. Lee Directors...... Maria Begonia Roy Duhé Juan Silva Administrative Assistant ...... Cynthia Huff

The Mississippi Academy of Sciences recognizes the following Gold Booth Exhibitor, 2005 Annual Meeting:

Base Pair Dr. Robin Rockhold University of Mississippi Medical Center 2500 North State St. Jackson, MS 39216-4505 601-984-1634 (phone) [email protected]

The Mississippi Center for Supercomputing Research (MCSR) provides free, high performance computing cycles and consulting in support of research and instruction, for all interested students, faculty, or researchers associated with any of Mississippi’s eight publicly funded institutions of higher learning. The MCSR actively supports the Mississippi Academy of Sciences with regular participation in the Mathematics, Computer Science, and Statistics Division. Please visit http://www.mcsr.olemiss.edu, email [email protected], or call 662-915-3922 to inquire about how we might support your HPC research or instructional computing projects at your university. Or, simply apply for an account today at http://www.mcsr.olemiss.edu/accounts.

164 Journal of the Mississippi Academy of Sciences The Anthropic Principle, String Theory, and Multiple Universes in Light of the Scientific Method

Jerome Goddard1 Mississippi Department of Health, Jackson, MS 39215

Epistemology is the study of the nature and is? Why is it different from that expected based upon grounds of knowledge, especially with reference to mathematical formulas? It is at this point that people its limits and validity. There are two main goals of invoke the anthropic principle. epistemology: (1) to find out as much truth as The anthropic principle is often used as a reli- possible, and (2) to avoid as much falsehood as gious argument for special creation with reasoning possible. These two goals are in tension with each like this, “The reason our universe is so peculiar and other. In trying to understand the world, the ancient well-fitted to life is because the Creator wanted Greek philosophers diligently strove to develop a (willed) it to be that way for the formation of life.” coherent and well-defined philosophy. Even though Dr. Steven Weinberg, a Nobel laureate from the modern technology and scientific advancements University of Texas, puts it like this. “A person is seem to render the ancient Greeks silly, supersti- dealt a royal flush in a poker tournament. It may be tious, or unenlightened, we are actually faced with chance, but on the other hand, the organizer of the the same challenge they were – to find a rational tournament may be our friend” (Overbye, 2003; explanation of the universe. Susskind, 2003). There has been considerable debate lately in the One way to get around the anthropic principle is physics community over an idea called the “anthro- to theorize that there have been millions of possible pic cosmological principle” (Barrow and Tipler, universes throughout time, with many different 1986; Barrow, 2002; Susskind, 2003). According to constants and settings in each one ruled by chance, this idea, the universe is made just right for life to and/or that there are “regions” (bubbles) within our occur. As one author puts it, “The universe must be universe containing a variety of constants and set- suitable for life, otherwise we would not be here to tings. The reason we are here to observe this universe wonder about it” (Overbye, 2003). There are numer- is because we happen to be in one of those multiple ous features and mathematical constants in the universes or perhaps bubbles within our universe equations of physics and cosmology which do not conducive to formation of life. String theory provides seem predictable by any known theory, and which some evidence to support this view. seem to be miraculously tuned to allow life. Any String theory. String theory is the idea that slight deviation from these settings would be disas- fundamental particles of the universe – protons, trous, causing things like stars to collapse and atoms neutrons, electrons, quarks, etc. – contain tiny vibrat- to evaporate. One of the most striking examples of ing, oscillating, dancing filaments called “strings.” the anthropic principle is the cosmological constant, String theory adds a new microscopic layer (a vibrat- a number that measures the amount of cosmic ing loop) to the previously known progression (large repulsion caused by the energy in empty space to small) from atoms to protons, neutrons, electrons, (Carroll and Press 1992). As predicted by quantum and quarks (Greene 2003; Greene 2004) . No one has theory, empty space should be brimming with this actually seen these strings; they are all theoretical at energy. In fact, recent discoveries have shown that this point. According to the idea, everything in the cosmic repulsion is indeed present and seems to be universe is comprised of tiny vibrating fundamental helping in the expansion of the universe. However, strings. Moreover, every one of these strings is the observed cosmological constant (lambda) is identical. The only difference between one string and perhaps as much as 1,000 times lower than its another, whether it is a heavy particle part of an atom estimated value (Weinberg, 1989; Carroll and Press, or a massless particle that carries light, is its resonant 1992). So why is the cosmological constant what it pattern, or how it vibrates (Greene, 2003).

1PO Box 1700, Jackson, MS 39215 601-576-7689; [email protected]

July 2005 Vol 50, No. 3 165 String theory helps resolve the incompatibility theory. One single observation of a black swan between quantum mechanics and general relativity allows us to logically derive the statement, “Not all – the properties of forces vs. the properties of parti- swans are white.” In this way of looking at things, cles. For this reason, many physicists think string empirical generalizations, though not verifiable, are theory can become a “theory of everything” because falsifiable. Under this new system, we can never it can underlie all others. A theory of everything – prove that what we know is true; it is always possible the ultimate explanation of the universe at its most that it will turn out to be false. The most you can ever microscopic level – would provide the firmest say about a theory – be it gravity, Big Bang, evolu- foundation on which to build our understanding of tion, etc – is that it is supported by every observation the world. so far, and yields more precise predictions than any String theory allows for many dimensions (10 or known alternative. However, it may be expanded or 11; perhaps even more) in a space framework which engulfed later by a better theory. For example, Ein- could serve as a landscape containing peaks, valleys, stein’s theory of special relativity expands Newto- and ridges (de Boer, 2003). As the universe ex- nian mechanics (NM) to high velocities, but NM is panded after the Big Bang, it theoretically rolled still used. down, or over, this landscape framework. By A critical premise in Popper’s method is that a chance, in some of these valleys small regions or theory must be testable. The hypothesis must lend pockets of universes (bubbles) might occur in which itself to experiments or observations in efforts to the natural constants and settings are conducive to disprove it. Taking the swan example (above), we life. must be able to go out and look at swans. If our theory about the color of swans relates to the color of MULTIPLE UNIVERSES? swans in another universe, then that theory is untest- able and thus lies outside the boundaries of science. Scientific method. To accurately understand our We can discuss it, yes, but test it, no. world, scientists developed the scientific method. Multiple universes or pockets within our uni- For a long time, the preferred scientific method was verse. As mentioned, one way to get around the induction – basing general statements on accumu- anthropic principle is to theorize that there have been lated observations of specific instances. For exam- millions of possible universes, with many different ple, “Since the sun has arisen every day during constants and settings ruled by chance. Alternatively, recorded history, it is therefore true that the sun will string theory might allow for pocket universes or rise tomorrow.” Use of induction was helpful and bubbles within our universe with a wide range of enabled early scientists to distinguish science from physical laws and constants. Accordingly, some non-science. However, David Hume in the mid- small fraction of those bubbles would be within the 1700s pointed out that no number of singular obser- anthropic window. And it is in one of these regions vations, however large, could logically fix a general that we find ourselves (Susskind, 2003). Again, statement or law (Melchert, 1999). His skepticism according to string theory, the only reason we are may have seemed silly to laypeople, but actually here to observe this universe or pocket within our unmasked a truth – pure empiricism is not a suffi- universe is because we happen to be in one of them cient basis for science. We do not know for sure that conducive to formation of life. the sun will rise tomorrow. It is highly probable that To me, proposing that ours is but one of a string it will (in light of historical observations), but we do of universes is a fallacy in reasoning. The universe is not know for sure. supposedly “everything there is.” That is what the A major shift in the scientific method occurred term universe means. If one tries to argue about in the 1900s when Karl Popper advanced the idea things outside the universe or before the universe, an that science should advance by trying to “falsify” opponent usually quips, “that’s outside the boundary hypotheses. Once falsified, they can either be of science. What lies outside the universe is an amended, qualified, or adjusted and the process unknowable nothingness.” Okay. Then how can an started over again. For example, the hypothesis, “all explanation for our universe being uniquely adapted swans are white” can be supported by thousands of for life be that ours is just one of many universes? observations of white swans. But it is really not There is no way to test such a statement. important how many observations agree with the The bubble idea suffers from the same flaw. Yes, theory; what really matters is trying to disprove the string theory math might allow for such phenomena,

166 Journal of the Mississippi Academy of Sciences but there is no way to test such a claim. Dr. Leonard ACKNOWLEDGMENTS Susskind at Stanford says, “Direct observational confirmation of the vastness and diversity of the Alan Penman, M.D., and T.J. Brooks, M.D., my friends (universe) landscape is probably not possible. The and fellow amateur philosophers, provided helpful comments space between bubbles is typically expanding so in the development of this manuscript. rapidly that no signals can reach one of them from LITERATURE CITED any other” (Susskind, 2003). Recall Popper’s method of hypothesis testing Barrow, J. D. 2002. The Constants of Nature. Pantheon Books, (more accurately “falsifying”). In the case of the New York, 352 pp. swans we can never say that all swans are white by Barrow, J. D., and F. J. Tipler. 1986. The Anthropic Cosmologi- counting white swans. But finding one black swan cal Principle. Oxford University Press, London, 706 pp. Carroll, S. M., and W. H. Press. 1992. The cosmological would answer the question. However, to test the constant. Ann. Rev. Astron. Astrophys. 30:499–542. hypothesis, we must be able to look at swans. There de Boer, J. 2003. String theory – an update. Nucl. Phys. Proc. is no possible way to test the multiple universe Suppl. 117:353–372. hypothesis – or even the pocket-universes-within- Greene, B. 2003. A theory of everything. Public Broadcasting our-universe hypothesis. As silly as it sounds, all Service, NOVA, “The Elegant Universe.” www.pbs.org/wgbh/nova/elegant/everything.html there is to observe is what there is to observe. Yet Greene, B. 2004. The Fabric of the Cosmos. Knopf Publishers, the multiple universe response is often quoted when New York, 569 pp. people discuss the anthropic principle. In my view, Melchert, N. 1999. The Great Conversation – A Historical we would do well to learn from the ancient Greek Introduction to Philosophy. Mayfield Publishing Co., Mountain View, CA, pp. 404–416. philosophers how to keep our theory of knowledge Overbye, D. 2003. Zillions of universes? Or did ours get lucky? consistent and coherent. New York Times, New York, October 28, 2003. Susskind, L. 2003. A universe like no other, pp. 34-41, 1 November, New Scientist. Weinberg, S. 1989. The cosmological constant problem. Rev. Modern Physics 61:1–23.

July 2005 Vol 50, No. 3 167 Red-cockaded Woodpecker (Picoides borealis) Behavior in a Mississippi Loblolly-Shortleaf Pine Forest

Douglas R. Wood1,2, L. Wesley Burger, Jr.2, and Francisco J. Vilella3

2Department of Wildlife and Fisheries, Mississippi State University, Mississippi State, MS 39762, and 3Mississippi Cooperative Fish and Wildlife Research Unit, Mississippi State, MS 39762

From 1997 to 1999, we characterized red-cockaded woodpecker (Picoides borealis) behavior in a loblolly (Pinus taeda) and shortleaf (P. echinata) pine forest in Mississippi. We recorded stem type and size class used, foraging location, height, first behavior type observed and cumulative behaviors during 5-hour visual observation periods of 41 red-cockaded woodpecker groups. Overall, 94% of all stems used by red-cockaded woodpeckers were pines, whereas only 6% of stems used were hardwoods. However, use of hardwood stems increased during the nonbreeding season. Red- cockaded woodpeckers selected large pine stems ( = 47.5 cm) compared to hardwood stems ( = 33 cm). During all seasons, red-cockaded woodpeckers foraged predominantly within the crown and high-trunk area of trees. Foraging and group cohesion behaviors were performed during all seasons, however foraging behaviors increased during the nonbreeding season.

Red-cockaded woodpeckers (Picoides borealis) from puddles or metabolically prey (Ligon, 1970; (RCW) are a federally-endangered species endemic Jackson, 1994). to mature longleaf (Pinus palustris), loblolly (P. RCWs frequently use large diameter pine trees taeda), and shortleaf (P. echinata) pine forests of the (Jackson, 1994); however, Ramey (1980) docu- southeastern United States (Jackson, 1994). RCWs mented greater percentages of hardwoods selected by are bark-probing insectivores that exhibit sexual foraging RCWs in Mississippi and South Carolina. segregation during foraging (Ligon, 1970; Jackson, Although RCWs prefer pines for foraging rather than 1994); however, most studies have been conducted hardwoods (Jackson, 1994), landscape and regional in longleaf pine forests (Morse, 1972; Engstrom and context may play a role in the frequency of hardwood Sanders, 1997). RCWs frequently use a foraging selection. Subpopulations in the eastern portion of strategy known as scaling or flaking to obtain prey the RCWs range inhabit longleaf pine ecosystems (Ligon, 1968). Feet or bills are used to remove large with low densities of hardwoods, typically found in flakes of bark to reveal prey. Ligon (1968, 1970) riparian areas or stream-side management zones. also reported flycatching and foliage gleaning by However, RCW subpopulations in the central and RCWs. western portion of the species’ range inhabit loblolly Adults typically forage on ants (Crematogaster and shortleaf pine ecosystems with greater densities and Camponotus spp.), spiders (Araneae), wood of hardwoods. These areas frequently have increased roaches (Parcoblatta spp.), beetle larvae and other hardwood densities throughout the landscape that are invertebrates (Beal, 1911; McFarlane, 1995; Hess not restricted to riparian areas. and James, 1998). RCWs may shift their diet to This research was conducted as part of a broader include more larvae during winter (Hess and James, study of RCW foraging ecology and reproductive 1998). In South Carolina, RCW nestlings were fed success (Wood, 2001). Our objectives were to char- larvae, wood roaches, spiders, ants and centi- acterize RCW foraging behavior, including stem use pedes (Scolopendromorpa spp.) (Harlow and Len- and type, relative location and height selection, first nartz, 1977; Hanula and Franzreb, 1995). RCWs also behavior type, and cumulative behaviors in a loblolly have been documented foraging in slash piles on the and shortleaf pine forest. ground (Ligon, 1970). RCWs obtain water directly

1Author for correspondence: Department of Biological Sciences, Southeastern Oklahoma State University, 1405 N. 4th Ave., PMB 4068, Durant, Oklahoma 74701–0609; [email protected]

168 Journal of the Mississippi Academy of Sciences MATERIALS AND METHODS We recorded the species of tree and the diameter at breast height (dbh) of each stem used for foraging. Research was conducted at the Bienville National We also recorded an individual RCW’s location on Forest (BNF) in central Mississippi. BNF consists of the tree within one of three relative location catego- 72,216 ha of pine, pine-hardwood, and hardwood ries (i.e., trunk within the crown, trunk below the stands in a fragmented landscape (Wood, 2001). crown, or limbs) (Ligon, 1968). Each location on the Dominant tree species include loblolly, longleaf, and tree was assigned a relative height category (i.e., shortleaf pine. Common hardwood species include lower third, middle third or higher third) within white oak (Quercus alba), post oak (Q. stellata), location categories (Wood, 2001). Thus, each loca- southern red oak (Q. falcata), mockernut hickory tion was assigned to a location-height category (e.g., (Carya tomentosa), sweetgum (Liquidambar styraci- low-crown or high-trunk). flua), and winged elm (Ulmus alata). We recorded the cumulative number and type of From 1997 to 1999, we characterized foraging behaviors (Table 1) performed during 1-min observa- and other behaviors of 41 RCW groups. Jackson tion periods (Kilham, 1959; Ligon, 1970; Kilham, (1994) defined an RCW group as the brood pair and 1974). We also recorded the behavior performed at $1 related RCWs which are usually sons of the the first second of 1-min observation periods, hereaf- brood pair. Each year, 12–15 RCW groups were ter referred to as first behavior, that provided an randomly-selected without replacement for intensive independent sample of behavior compared to cumu- monitoring from the population of 95 active groups lative behavior counts. First behavior data were only at BNF. Overall, approximately 1925 hours of obser- collected in 1998 and 1999 as a modification from vations were recorded for 41 different RCW groups. the original protocol performed in 1997 (Wood, Five-hour visual observation periods were performed 2001). on each group for one year beginning in January and We used SAS version 7.0 for all statistical concluding in December. Observation periods were analyses (" = 0.05) (SAS Institute 1998). We tested performed daily and sequentially by group through- the hypothesis that the number of pine and hardwood out the year to approximate equal effort. Each stems used by foraging RCWs was equal among observation period began at first light and continued groups and seasons. We used a log-linear analysis to for five hours (Engstrom and Sanders, 1997). The examine the 3-way interaction terms of group, breeding season was defined as 7 April–31 July and season, and stem type. Chi-square tests were used to the nonbreeding season as 1 January–6 April and 1 test for differences between seasons and stem types, August–15 December (Jackson, 1994). Each obser- as well as differences between groups and stem vation period was subdivided into 6-min periods types. We also tested the hypothesis that dbh of pine consisting of a 1-min observation period followed by and hardwood stems used by foraging RCWs was a 5-min waiting period when no data were collected equal. Within pine and hardwood classes, we tested (Hooper et al., 1982; DeLotelle et al., 1983). During the hypothesis that there were no differences in dbh the 5-hour observation period, as many 6-min peri- between seasons. We used unbalanced, mixed-model ods as possible were recorded (Brennan and Morri- ANOVAs to test for effects of group, year, season, son, 1990). and stem type on dbh of stems used by foraging RCWs forage as a group, thus we defined the RCWs. group as the sampling unit for characterizing behav- Chi-square tests of homogeneity were used to test iors (Sherrill and Case, 1980; Hooper et al., 1982; the null hypothesis that RCWs used similar location- Doster and James, 1998). Individuals were classified height categories on foraging stems annually and by age (e.g., juvenile males can be identified by a red during the breeding and nonbreeding season. We patch of feathers on the crown of the bird) and sex used equal prior probabilities because previous when possible, however identification proved diffi- studies conflicted on RCW foraging locations (Ligon, cult due to dense midstory vegetation and we were 1970; Morse, 1972). Because our study design was not able to color-mark all RCWs. During 1-min observational, we made no attempt to test location- observation periods, one RCW from the group was height observations with availability of these catego- observed and all locations of that individual were ries among various stem types used. For first behav- spot-mapped on graph paper and subsequently each ior analysis, we used PROC GENMOD on the actual RCW location was georeferenced with a differen- counts of first behavior for all RCW groups com- tially-corrected global positioning system unit. bined annually and during the breeding and non-

July 2005 Vol 50, No. 3 169 Table 1. Red-cockaded woodpecker (Picoides borealis) behaviors observed at Bienville National Forest, Mississippi 1997–1999. Behavior Definition Foraging Actively capturing prey, scaling, flaking, probing Hunting Actively searching a substrate Calling Vocalizing Loafing Motionless on a substrate Preening Arranging or smoothing plumage Cavity maintenance Drilling resin wells near cavity; not excavating inside a cavity Drumming Tapping on substrate for communication; not drilling resin wells Fed nestling Providing food to a nestling Fed fledgling Providing food to a recently-fledged offspring Cavity excavation Actively excavating inside a cavity Wing display Raised wings above body Scratching Using toes to scratch Play/social interaction Intragroup chases, territorial behavior, dominance behaviors

P 2 breeding season. If a year effect was detected, years seasons ( 1 = 38.7, P < 0.001) and among groups P 2 were analyzed separately. ( 40 = 242.5, P < 0.001). Stem diameter. Foraging RCWs frequently used RESULTS large diameter pines and, less frequently, large Stem use. Loblolly and shortleaf pine accounted diameter hardwoods (Table 4). No year effect was for 99% of all foraging observations on pines, detected (F2,38 = 2.03, P = 0.15), however dbh of whereas post oak, white oak, winged elm, southern pines was greater than hardwoods (F1,9703 = 770.09, P red oak, and sweetgum accounted for 82% of hard- < 0.001) for the breeding and nonbreeding seasons. woods selected by foraging RCWs (Table 2). RCWs A difference in dbh between seasons (F1,9703 = 27.47, anecdotally were observed foraging on eight tree P < 0.001) was detected for pines and hardwoods. species (spruce pine [P. glabra], American beech Location-height. Annually, RCWs were ob- [Fagus grandifolia], black cherry [Prunus serotina], served more frequently on the trunk and limbs within P 2 blackjack oak [Q. marilandica], pignut hickory the crown than on the trunk below the first limb ( 1 [Carya glabra], sassafras [Sassafras albidum], = 2895, P < 0.001) (Table 5). RCWs were more sugarberry [Celtis laevigata], and swamp chestnut frequently observed on the trunk within the crown P 2 oak [Q. michauxii]), but were not recorded during 1- than on the trunk below the crown and on limbs ( 2 min observation periods. = 3098, P < 0.001). In ranked order, RCWs foraged Annually and during the breeding season, RCWs more frequently in the mid-crown, low-crown, high- P 2 foraged more frequently on pines than hardwoods crown, and high-trunk areas of foraging stems ( 8 = (Table 3). However, we observed an increase in 5016, P < 0.001) (Table 5). hardwood use compared to pines during the During the breeding season, RCWs were ob- nonbreeding season (Table 3). RCWs foraged more served more frequently on the trunk and limbs within P 2 frequently on hardwoods during the nonbreeding the crown than on the trunk below the first limb ( 1 P 2 season than the breeding season ( 1 = 31.3, P < = 408, P < 0.001). RCWs also were more frequently 0.001). Stem type use by RCWs varied between observed on the trunk within the crown than on the

170 Journal of the Mississippi Academy of Sciences P 2 trunk below the crown and on limbs ( 2 = 636, P < 0.001). RCWs foraged more frequently on the high-trunk and mid- P 2 crown areas of trees than other areas ( 8 = 1627, P < 0.001) (Table 5). Table 2. Number and percent of pine and hardwood During the nonbreeding season, species used by foraging red-cockaded woodpeckers RCWs were observed more frequently on (Picoides borealis) at Bienville National Forest, the trunk within the crown and limbs Mississippi 1997–1999. within the crown than on the trunk below P 2 Stem type Numbe % the first limb ( 1 = 2959, P < 0.001). RCWs also were observed more fre- r quently on the trunk within the crown Pine than on the trunk below the crown and on P 2 Loblolly Pine (Pinus taeda) 8082 88.0 limbs ( 2 = 3183, P < 0.001). RCWs foraged more frequently in the mid-crown Shortleaf Pine (P. echinata) 1005 11.0 P 2 and high-crown areas of trees ( 8 = 4371, P < 0.001) (Table 5). Pine Snag (Pinus spp.) 67 0.5 First behavior. A year by season Longleaf Pine (P. palustris) 50 0.5 P 2 interaction effect ( 1 = 55.0, P < 0.001) was detected for first behavior, thus years Total 9204 100.0 were analyzed separately (Table 6). In Hardwood 1998 and 1999, preening, cavity mainte- Post Oak (Quercus stellata) 141 26.0 nance, drumming, fed nestling, fed fledg- ling, excavation, wing display, scratching, White Oak (Q. alba) 92 17.0 and play behaviors were not performed Winged Elm (Ulmus alata) 91 17.0 frequently enough to permit meaningful analysis (Table 6). During 1998, foraging Southern Red Oak (Q. falcata) 70 12.0 P22P (11 = 183.5, P < 0.001), hunting ( = Sweetgum (Liquidambar styraciflua) 56 10.0 P 2 98.9, P < 0.001), calling ( 1 = 9.8, P = P 2 Mockernut Hickory (Carya tomentosa) 28 5.0 0.002), and loafing ( 1 = 23.8, P < 0.001) were performed more during the non- Hardwood Snag 23 4.0 breeding season than the breeding season, however variation existed among groups Northern Red Oak (Q. rubra) 17 3.0 P 2 for foraging ( 13 = 175.8, P < 0.001), Water Oak (Q. nigra) 7 1.0 P 2 hunting ( 13 = 90.5, P < 0.001), calling P 22P Willow Oak (Q. phellos) 7 1.0 ( 13 = 24.0, P = 0.03), and loafing ( 13 = 27.4, P = 0.011). During 1999, foraging Black Gum (Nyssa sylvatica) 6 1.0 P22P (11 = 0.23, P = 0.63), calling ( = 0.62, P 2 Flowering Dogwood (Cornus florida) 2 0.5 P = 0.43), and loafing ( 1 = 0.88, P = 0.35) were not different between seasons, Shagbark Hickory (Carya ovata) 2 0.5 however variation was detected among P 2 White Ash (Fraxinus americana) 1 0.5 groups for foraging ( 11 = 59.9, P < P 2 0.001), calling ( 11 = 31.7, P = 0.001), Red Maple (Acer rubrum) 1 0.5 P 2 and loafing ( 11 = 25.0, P = 0.01). P 2 Yellow Poplar (Liriodendron tulipifera) 1 0.5 Hunting ( 1 = 72.2, P < 0.001) was performed more frequently during the Total 545 100.0 nonbreeding season than the breeding season, however variation was detected P 2 among groups ( 11 = 72.7, P < 0.001).

July 2005 Vol 50, No. 3 171 Table 3. Number, mean percent, and range of percentage of stem type used by red-cockaded woodpeckers (Picoides borealis) at Bienville National Forest, Mississippi 1997–1999. Season Stem type Number % Range % of stems Pine 9204 94 83–100 Annual Hardwood 545 6 0–17 Pine 4459 96 86–100 Breeding Season Hardwood 187 4 0–14 Pine 4742 93 69–100 Nonbreeding Season Hardwood 361 7 0–31

Table 4. Mean (± SE) dbh (cm) of pine and hardwood stems used annually and seasonally by foraging red-cockaded woodpeckers (Piciodes borealis) at Bienville National Forest, Mississippi 1997–1999. Year Season Type Number 0 ± SE of stems 1997 Pine 1737 48.3 ± 0.3 Breeding Hardwood 99 35.1 ± 1.6 Pine 1131 46.5 ± 0.4 Nonbreeding Hardwood 151 32.3 ± 0.9 1998 Pine 1739 45.2 ± 0.3 Breeding Hardwood 72 30.0 ± 1.7 Pine 2438 47.5 ± 0.3 Nonbreeding Hardwood 185 31.0 ± 1.1 1999 Pine 983 47.2 ± 0.4 Breeding Hardwood 16 33.0 ± 3.6 Pine 1173 51.5 ± 0.3 Nonbreeding Hardwood 25 43.5 ± 2.1 1997–1999 Pine 4459 47.0 ± 0.3 Breeding Hardwood 187 33.0 ± 0.8 Pine 4742 48.3 ± 0.3 Nonbreeding Hardwood 361 32.5 ± 1.0

172 Journal of the Mississippi Academy of Sciences We observed some of the highest rates of Table 5. Percent of red-cockaded woodpecker hardwood stem selection by RCW groups, partic- (Picoides borealis) observations (n = 11,165) ularly during the nonbreeding season, compared by location-height categories annually and to the literature (Skorupa and McFarlane, 1976; Ramey, 1980). For example, 31% of all stems seasonally at Bienville National Forest, used by one RCW group were hardwoods in Mississippi 1997–1999. 1999. BNF, in the West Gulf Coastal Plain, has an Breeding Nonbreeding Annual increased hardwood component compared to season season longleaf systems in the southeastern portions of the RCWs’ range (Rudolph and Conner, 1996). Crown-high 11.7 18.4 15.3 Thus, more hardwoods were potentially available Crown-mid 20.0 32.3 26.7 in the overstory for foraging activities. Increased hardwood use during the nonbreeding season also Crown-low 14.7 16.7 15.8 may be due to social dominance. Jones and Hunt Trunk-high 21.4 10.3 15.4 (1996) suggested dominance and sexual segrega- tion may pressure juvenile RCWs lower on pines Trunk-mid 11.0 3.6 6.9 or onto nearby hardwoods. We also observed Trunk-low 3.4 1.2 2.2 juvenile male RCWs foraging on smaller dbh hardwoods such as winged elm and sweetgum. Limbs-high 3.3 5.2 4.3 Skorupa and McFarlane (1976) reported that Limbs-mid 9.2 9.5 9.5 RCWs in South Carolina did not forage on hard- woods in summer. However, 10% of all RCW Limbs-low 5.3 2.8 3.9 foraging stems in winter were hardwoods; sug- gesting that decreasing winter prey availability in pine stands may increase the use of hardwoods by RCWs (Skorupa and McFarlane, 1976). In Flo- DISCUSSION rida, DeLotelle et al. (1987) also reported a seasonal increase in the use of hardwood stems by At BNF, RCWs frequently used pines for forag- RCWs. They foraged on baldcypress (Taxodium ing activities. Zwicker and Walters (1999) reported distichum) in greater proportion than availability that 94% of all RCW foraging stems were pines and during the nonbreeding season (DeLotelle et al., were used in greater proportion than availability. In 1987). Louisiana, RCWs selected pines greater than their Hardwood species selected by RCWs at BNF, availability (90% use; 64% available), whereas hard- such as white oak, post oak and southern red oak, woods were selected less than their availability have deeply-grooved or loose bark that may appear (Jones and Hunt, 1996). During the breeding season, texturally similar to pine substrates used by RCWs. RCWs in Louisiana foraged more frequently on An alternate hypothesis may be that RCWs forage on smaller pine stems (< 40 cm dbh) than during the hardwoods due to increased invertebrate availability nonbreeding season (Jones and Hunt, 1996). compared to pine stems. Hardwoods may harbor In Arkansas, Doster and James (1998) docu- more invertebrates than pines during winter months, mented 95% foraging on shortleaf pines compared especially after invertebrates have been depleted to 5% hardwoods. In Mississippi and South Caro- from nearby pines by a central-place forager like the lina, Ramey (1980) reported 78–94% foraging on RCW (Skorupa and McFarlane, 1976; DeLotelle et pines. However, higher rates of foraging on pines is al., 1987). However, other bark-probing woodpecker frequently reported from longleaf forests which are species may exclude RCWs from hardwoods. During more homogenous in terms of composition. In our study, we observed red-bellied woodpeckers Florida, Porter and Labisky (1986) reported 99% (Melanerpes carolinus), red-headed woodpeckers foraging on pine stems and Hardesty et al. (1997) (M. erythrocephalus), and downy woodpeckers (P. reported 97% of stems used by foraging RCWs were pubescens) displace RCWs from hardwoods. pines.

July 2005 Vol 50, No. 3 173 Table 6. Mean number (± SE) of first behaviors of red-cocka- ded woodpecker (Picoides borealis) groups annually and sea- sonally at Bienville National Forest, Mississippi 1998–1999. Year Behavior Breeding Non- Annual season breeding season 1998 Foraging 50.1 ± 5.2 93.1 ± 9.2 71.6 ± 6.6 Hunting 43.4 ± 4.3 71.7 ± 5.7 57.5 ± 4.5 Calling 9.5 ± 1.2 13.5 ± 1.5 11.5 ± 1.0 Loafing 11.0 ± 1.6 5.71 ± 1.0 8.36 ± 1.1 Preening 2.86 ± 0.6 1.0 ± 0.3 1.93 ± 0.4 Cavity mainte- 2.57 ± 0.8 0.86 ± 0.4 1.71 ± 0.5 nance Drumming 0.21 ± 0.1 0.0 0.11 ± 0.1 Fed nestling 0.43 ± 0.4 0.0 0.21 ± 0.2 Fed fledgling 0.5 ± 0.4 0.0 0.25 ± 0.2 Cavity excavation 6.1 ± 3.0 0.0 3.04 ± 1.6 Wing display 0.07 ± 0.1 0.0 0.04 ± 0.1 Scratching 0.07 ± 0.1 0.0 0.04 ± 0.1 Play 0.57 ± 0.2 0.0 0.29 ± 0.1 1999 Foraging 39.9 ± 3.3 41.2 ± 5.1 40.5 ± 3.0 Hunting 18.8 ± 2.6 37.0 ± 4.6 27.9 ± 3.2 Calling 27.9 ± 1.0 26.3 ± 3.1 27.1 ± 1.6 Loafing 6.17 ± 1.1 5.25 ± 0.8 5.71 ± 0.7 Preening 0.75 ± 0.3 0.42 ± 0.2 0.58 ± 0.2 Cavity mainte- 1.17 ± 0.6 1.17 ± 0.9 1.17 ± 0.5 nance Drumming 0.08 ± 0.1 1.42 ± 0.8 0.75 ± 0.4 Fed nestling 2.33 ± 0.6 0.0 1.17 ± 0.4 Fed fledgling 0.67 ± 0.3 0.0 0.33 ± 0.2 Cavity excavation 1.92 ± 0.8 0.0 0.96 ± 0.5 Wing display 0.25 ± 0.2 0.08 ± 0.1 0.17 ± 0.1 Scratching 0.0 0.0 0.0 Play 0.0 0.08 ± 0.1 0.04 ± 0.1

174 Journal of the Mississippi Academy of Sciences Stem diameter. Mean dbh of pines used by groups with > 2 members, thus the increase in forag- RCWs at BNF were similar to mean dbhs reported in ing observations on the trunk above the first limbs other loblolly-shortleaf pine forests. In a mixed and on limbs. Ramey (1980) documented that males Louisiana forest, > 50% of pines selected by forag- and females preferred foraging on trunks, but males ing RCWs ranged from 40–60 cm dbh, which was foraged more on limbs than females in a Mississippi greater than their availability (Jones and Hunt, loblolly pine forest. In an old-growth longleaf pine 1996). Further, pines < 40 cm dbh were selected forest, Engstrom and Sanders (1997) documented a more during the breeding season than the nonbreed- similar pattern of resource partitioning in RCWs: ing season (Jones and Hunt, 1996). In an Arkansas males foraged more on the upper trunk and limbs, shortleaf forest, Doster and James (1998) docu- whereas females foraged on the trunk below the first mented that 75% of all RCW foraging stems were limbs. RCWs segregated further by foraging sub- $ 30.5 cm dbh, although no stems > 38 cm dbh were strate as well, with females concentrating on bark- available on their site. Doster and James (1998) also probing whereas males used a diversity of food observed RCWs foraging on stems as small as sources (Engstrom and Sanders, 1997). 7.6–15.2 cm dbh. Behavior. RCWs at BNF allocated the majority Hardesty et al. (1997) reported a mean dbh of of their behaviors to foraging and group cohesion 29.7 cm for stems selected by RCWs in a Florida maintenance (e.g., calling). The trend towards in- longleaf pine forest. They reported a wide range of creased time spent foraging was more evident during longleaf pine size classes, similar to BNF, used by the nonbreeding season when food availability RCWs (3.2–72.4 cm). In a North Carolina longleaf declines and the need for thermoregulation increases forest, Zwicker and Walters (1999) reported that the with decreased ambient temperatures. During the majority of stems selected by RCWs ranged from breeding season, RCWs at BNF performed more 20.1–35.0 cm. Similarly, pines $ 25.1 cm dbh were behaviors related to reproductive activities such as used more than their availability and 5.1–25.0 cm cavity excavation, copulation, and feeding offspring. dbh pines were used less than their availability We observed several interesting foraging behav- (Zwicker and Walters, 1999). iors during our study. In the summer of 1998, RCWs In an old-growth longleaf pine forest, Engstrom took advantage of a cicada outbreak. RCWs would and Sanders (1997) reported that 80% of pines used sally from a tree, catch a cicada in flight, return to the by RCWs ranged from 35–65 cm. Pines >31 cm same tree and ingest the insect. On several occasions, were used greater than their availability, whereas we observed RCWs foraging for on deadfall stems < 20 cm were used less than their availability and drinking from puddles on the ground similar to (Engstrom and Sanders, 1997). In a Florida longleaf observations by Ligon (1970) and Jackson (1994). forest, DeLotelle et al. (1983) reported that RCWs preferred pines $ 27 cm dbh. However, DeLotelle et ACKNOWLEDGMENTS al. (1983) also reported extensive use of smaller age classes (12–16 cm). Small pines accounted for 27% We thank J. Altman, W. Bailey, R. Carlson, R. Chambers, of the stems available, but incurred 31% use by E. Grant, R. Hamrick, D. Harrington, J. Kelly, C. McRae, M. Proett, C. Reynolds, S. Samano, S. Smith, T. Sullivan, K. RCWs. Thus, foraging on smaller dbh stems is not Swanson, M. Taquino, S. Tedford and N. Winstead for their limited to RCW subpopulations in loblolly-shortleaf dedicated efforts. R. Conner, R. Engstrom, D. Rudolph, and D. pine systems. Wilkins provided constructive comments on preliminary drafts. Location-height. We characterized general We are indebted to D. Elsen and the staff at the Bienville National Forest Ranger District for technical and administrative patterns in location and height by foraging RCWs, support. We thank the U. S. Forest Service, the Mississippi although we were unable to obtain enough data to Museum of Natural Science, and the Forest and Wildlife analyze differences between males and females. Research Center at Mississippi State University for financial Previous studies have documented sexual segrega- support of this study. tion and resource partitioning by RCWs in longleaf pine forests (Engstrom and Sanders 1997), however LITERATURE CITED we were unable to document similar results in a loblolly-shortleaf pine forest. Annually, and for both Beal, F.E.L. 1911. Food of the woodpeckers of the United States. U.S.D.A. Biological Survey Bulletin 37:22–23. seasons, RCWs foraged on the trunk above the first Brennan, L.A., and M.L. Morrison. 1990. Influence of sample limbs and on limbs. Our results may reflect the size on interpretations of foraging patterns by chestnut- presence of male helpers and juveniles in RCW backed chickadees. Studies in avian biology 13:187–192.

July 2005 Vol 50, No. 3 175 DeLotelle, R.S., J.R. Newman, and A.E. Jerauld. 1983. Habitat _____. 1974. Play in hairy, downy, and other woodpeckers. use by red-cockaded woodpeckers in central Florida. Pages Wilson Bulletin 86:35–42. 59–67 in D.A. Wood, ed. Red-cockaded Woodpecker Ligon, J.D. 1968. Sexual differences in foraging behavior in Symposium II. Florida Game and Freshwater Fish Com- two species of Dendrocopus woodpeckers. Auk mission, Tallahassee, Florida. 85:203–215. DeLotelle, R.S., R.J. Epting, and J.R. Newman. 1987. Habitat _____. 1970. Behavior and breeding biology of the red-cocka- use and territory characteristics of red-cockaded wood- ded woodpecker. Auk 87:255–278. peckers in central Florida. Wilson Bulletin 99:202–217. McFarlane, R.W. 1995. The relationship between body size, Doster, R.H., and D.A. James. 1998. Home range size and trophic position and foraging territory among woodpeckers. foraging habitat of red-cockaded woodpeckers in the Pages 303–308 in D.L. Kulhavy, R.G. Hooper, and R. Ouachita Mountains of Arkansas. Wilson Bulletin Costa, eds. Red-cockaded Woodpecker: recovery, ecology 110:110–117. and management. Center for Applied Studies in Forestry, Engstrom, R.T., and F.J. Sanders. 1997. Red-cockaded wood- College of Forestry, Stephen F. Austin State University, pecker foraging ecology in an old-growth longleaf pine Nacogdoches, Texas. forest. Wilson Bulletin 109:203–217. Morse, D.H. 1972. Habitat utilization of the red-cockaded Hanula, J.L., and K.E. Franzreb. 1995. prey of woodpecker during the winter. Auk 89:429–435. nestling red-cockaded woodpeckers in the upper coastal Porter, M.L., and R.F. Labisky. 1986. Home range and foraging plain of South Carolina. Wilson Bulletin 107:485–495. habitat of red-cockaded woodpeckers in northern Hardesty, J.L., K.E. Lucas, and H.F. Percival. 1997. Ecological Florida. Journal of Wildlife Management 50:239–247. correlates of red-cockaded woodpecker (Picoides borealis) Ramey, P. 1980. Seasonal, sexual, and geographical variation foraging preference, habitat use and home range size in in the foraging ecology of the red-cockaded woodpecker northwest Florida (Eglin Air Force Base). Florida Cooper- (Picoides borealis). Thesis, Mississippi State University, ative Fish and Wildlife Research Unit, University of Starkville, Mississippi, USA. 129 pp. Florida, Gainesville, Florida. Final Report. 80 pp. Rudolph, D.C., and R.N. Conner. 1996. Red-cockaded wood- Harlow, R.F., and M.R. Lennartz. 1977. Foods of nestling red- peckers and silvicultural practices: is uneven-aged silvicul- cockaded woodpeckers in coastal South Carolina. Auk ture preferable to even-aged? Wildlife Society Bulletin 94:376–377. 24:330–333. Hess, C.A., and F.C. James. 1998. Diet of the red-cockaded SAS Institute. 1998. SAS/STAT® user’s guide. Version 7. SAS woodpecker in the Apalachicola National Forest. Journal Institute, Cary, North Carolina, USA. of Wildlife Management 62:509–517. Sherrill, D.M., and V.M. Case. 1980. Winter home range of 4 Hooper, R.G., L.J. Niles, R.F. Harlow, and G.W. Wood. 1982. clans of red-cockaded woodpeckers in the Carolina Sand- Home ranges of red-cockaded woodpeckers in coastal hills. Wilson Bulletin 92:369–375. South Carolina. Auk 99:675–682. Skorupa, J.P., and R.W. McFarlane. 1976. Seasonal variation in Jackson, J.A. 1994. Red-cockaded Woodpecker. Pages 1–20 in foraging territory of red-cockaded woodpeckers. Wilson A. Poole and F.B. Gill, eds. The Birds of North America, Bulletin 88:662–665. no. 85. The Academy of Natural Sciences, Philadelphia; Wood, D.R. 2001. Multi-resolution assessment of the foraging The American Ornithologists’ Union, Washington, D.C. and reproductive ecology of red-cockaded woodpeckers in Jones, C.M., and H.E. Hunt. 1996. Foraging habitat of the red- a Mississippi loblolly-shortleaf pine forest. Dissertation, cockaded woodpecker on the D’Arbonne National Wildlife Mississippi State University, Starkville, Mississippi. 222 Refuge, Louisiana. Journal of Field Ornithology pp. 67:511–518. Zwicker, S.M., and J.R. Walters. 1999. Selection of pines for Kilham, L. 1959. Head-scratching and wing-stretching of foraging by red-cockaded woodpeckers. Journal of Wildlife woodpeckers. Auk 76:527–528. Management 63:843–852.

176 Journal of the Mississippi Academy of Sciences Consumer Preference of Apples Grown in Northern Mississippi1

Maria J. Sindoni2, Frank B. Matta3,4, and Juan L. Silva5

2Instituto National de Investigacion Agropecuaria (INIA)- Anzoátegui, Venezuela; 3Department of Plant and Soil Sciences, Mississippi State University, Mississippi State, MS 39762; and 5Department of Food Science and Technology, Mississippi State University, Mississippi State, MS 39762

Apples of cultivar/rootstock combination ‘Earligold’/EMLA 7, ‘Jonagold’/EMLA 111, ‘Improved Golden’/EMLA 7, ‘Improved Golden’/EMLA 111, ‘Scarlet Gala’/EMLA 7, ‘Jonafree’/Mark, ‘Macspur’/M7A, ‘Royal Gala’/MM 111, and ‘William’s Pride’/M7A were evaluated by an untrained consumer panel at harvest, 30 days, and 60 days after harvest. Changes in apple appearance, flavor, sweetness, tartness, and firmness were rated. All combinations except ‘Jonafree’/Mark and ‘Macspur’/M7A had a high acceptance rating by the panelists during the study. Fruit of ‘Earligold’/EMLA 7 and ‘William’s Pride’/M7A had a moderate acceptance by the panelists at harvest. Results indicated that fruit of medium and late cultivars such as ‘Improved Golden’/EMLA 7, and ‘Royal Gala’/ MM111 were preferred by panelists compared to fruit of the early harvested cultivars ‘Earligold’/EMLA 7 and ‘William’s Pride’/ M7A.

The consumer traditionally plays a major role in It is important to evaluate apple cultivar acceptance determining fruit acceptability for marketing of fresh both at harvest and after storage. Many scientists and stored fruit (Wills et al., 1998). Consumers’ have used taste panels to determine quality of apples perception of apple quality include such factors as and most studies are concerned with preferences or appearance, texture, firmness, sweetness, and flavor differences among cultivars (Watada et al., 1980). with preference and taste being key factors affecting Plotto et al. (1997) and Williams and Langron (1983) consumer purchase decisions. Growers can improve have used sensory science such as hedonic scales or product attributes, competitiveness, and marketabil- intensity scales to describe apple cultivars. Since the ity by using knowledge of consumers demand taste evaluations of Janson (1972), little has been (Brumfield et al., 1993). Watkins et al. (1993) published in North America on taste ratings of indicated that in ‘York Imperial,’ apple fruit firm- apples, especially the newer cultivars. In addition, ness and soluble solid contents (SSC) were the best there is little information on which parameters to use indicators of fruit maturity and quality. Optimal in measuring consumer preference of apples. Apples quality for Washington apples was obtained for fruit have many divergent attributes that are associated harvested 173 to 180 days after full bloom (Plotto et with acceptability and/or desirability (Watada et al., al., 1995). In addition to maintaining high quality 1980). Williams and Langron (1983) conducted a standards, good storage life is essential in successful study of attributes that panelists recognized in ‘Cox’s marketing and selling of fruit to consumers (Patte, Orange Pippin’ apples and concluded that quality of 1985). Researchers have found that apples can be apples can be characterized best by identifying the stored from -1 °C and 4 °C for 90 days while main- significant attributes, and then determining the taining quality (Westwood, 1993). Johnson and intensity of such attributes. The purpose of this study Ertan (1983) reported that ‘Idared’ apples stored at was to determine consumer preference for apples 1 °C were firmer than those kept at 0 °C or 4 °C. based on appearance, flavor, sweetness, tartness, and Shelf life after storage is also an important aspect of firmness of fruit at harvest, and after harvest storage. cultivar evaluations (Moore and Ballington, 1990).

1Approved for publication as Journal Article No. 10569 of the Mississippi Agricultural and Forestry Experiment Station, Mississippi State University.

4Author for Correspondence: Box 9555, 117 Dorman Hall; [email protected]

July 2005 Vol 50, No. 3 177 MATERIALS AND METHODS scale, where 1 = dislike extremely (very low), 5 = neither like or dislike (moderate), and 10 = like Trees producing fruit for the experiment were extremely (very high). seven years old and grown in Atwood silt loam soil A completely randomized design with repeated at the Pontotoc Ridge-Flatwoods Research and measures was used in the experiment. Data were Extension Center (38°08 N, 89°00' W), located seven analyzed using PROC GLM (SAS Statistical Soft- miles south of Pontotoc, MS. The average annual ware, SAS Institute, Gary, N.C.). Treatment means maximum temperature of this area is 30 °C (86 °F) were separated by LSD, 5% significance level. and minimum temperature is -1 °C (30 °F), with annual rainfall of 81.23 cm (32 inches). Research RESULTS AND DISCUSSION trees were planted in 1993 at a spacing of 2.5 m in rows and 3.7 m between rows. Trees were pruned to At harvest, consumer preference based on fruit a modified central leader system. The soil pH was appearance did not vary among cultivar/rootstock 5.6. In May 1999, a 5-20-20 fertilizer was applied at combinations, except for ‘Macspur’/M7A which was a rate of 450 g per tree, and ammonium nitrate (34-0- least preferred (Table 1). Thirty days after storage, 0) at a rate of 230 g per tree. No irrigation was consumer preference for ‘Earligold’/EMLA7 and applied. Weeds were controlled in the row by appli- ‘Macspur’/M7A, was least. Fruit appearance among cation of Round-up® herbicide in a one meter strip, the remaining cultivar/rootstock combinations did and a mowed strip was maintained between rows. not differ (Table 2). Sixty days after storage, ‘Jona- Insects and diseases were controlled through a spray free’/Mark and ‘Macspur’/M7A were least preferred program as recommended by Mississippi State and ‘Earligold’/EMLA7 and ‘William’s Pride’/M7A University Extension Service. had completely deteriorated, hence the sensorial test Fruit from cultivar/rootstock combinations for appearance of these cultivars was not possible. ‘Earligold’/EMLA 7, ‘Jonagold’/EMLA 111, ‘Im- Consumer preference based on appearance among proved Golden’/EMLA 7, ‘Improved Gol- the remaining cultivars did not differ and ranged den’/EMLA 111, ‘Scarlet Gala’/EMLA 7, ‘Jona- from 6.6 to 7.3 (Table 3). Similarities in preference free’/Mark, ‘Macspur’/M7A, ‘Royal Gala’/MM 111, among most of the cultivars suggest that panelists and ‘Williams Pride’/M7A were included in this were consistent in rating the apples. Kappel et al. study. The experiment consisted of four single tree (1992) reported that in a sensory evaluation, only the replications. Fruit quality and sensory evaluations visual attributes were found to be significantly were conducted immediately after harvest, 30 days, different among the strains of ‘Gala’ and ‘Jonagold.’ and 60 days after storage at 2 °C and 71% relative Fruit appearance will be influenced by the intended humidity (RH). Parameters evaluated were fruit size, market and this should be considered when selecting expressed as fruit length and diameter, soluble solids a cultivar for commercial production. Brumfield et content (SSC), juice pH, and firmness. Fruit size was al. (1993) reported similar conclusions when looking measured using a vernier caliper and fruit firmness at consumer tastes and preferences in purchasing was measured using a penetrometer (Instron Univer- fresh tomatoes. sal Machine, Model 1011, Canton, MA) and mea- At harvest, consumer preference based on flavor sured in Newtons. Juice soluble solids content was indicated a high preference for ‘Jona- measured in Brix with a Bausch & Lomb Abbe 3 L gold’/EMLA111, and ‘Improved Golden’/EMLA7. refractometer, and juice pH was measured using an These combinations were harvested mid and late Accumet pH meter 925 (Fisher Scientific, Pittsburgh, season. The least preferred by the panelists were PA). ‘Jonafree’/Mark and ‘Macspur’/M7A. Flavor did not Five apples from each tree were washed and cut differ among the remaining cultivar/rootstock combi- longitudinally and placed on paper plates for panel- nations (Table 1). Thirty days after storage, ‘Jona- ists to evaluate. A whole apple was also placed on gold’/EMLA111 and ‘Improved Golden’/EMLA7 the plate to be evaluated. Twenty-four people among maintained the highest preference rating by the students and staff were chosen at random from panelists and ‘Jonafree’/Mark and ‘Macspur’/M7A, Dorman Hall, Plant and Soil Science Department, at the lowest rating (Table 2). Sixty days after storage, Mississippi State University to participate in the test. ‘Earligold’/EMLA7 and ‘William’s Pride’/M7A Each panelist rated the apples for appearance, flavor, could not be tested due to fruit deterioration (Table sweetness, tartness, and firmness using a ten point 3). Similar results were reported by Plotto et al.

178 Journal of the Mississippi Academy of Sciences (1997) where distribution of sensory scores sug- spur’/M7A, and ‘William’s Pride’/M7A were least gested that early harvested fruit had not developed preferred (Table 2). Sixty days after storage, the full flavor and that high values for tartness and highest preference rating were for ‘Improved Gol- firmness did not necessarily imply quality and den’/EMLA7 followed by ‘Jonafree’/Mark, ‘Jona- consumer acceptance. gold’/EMLA111, ‘Royal Gala’/MM111 and ‘Scarlet At harvest, consumer preference based on sweet- Gala’/EMLA7, while ‘Improved Golden’/EMLA111, ness indicated that ‘Earligold’/EMLA7, ‘Jona- ‘Jonafree’/Mark, and ‘Macspur’/M7A were least free’/Mark and ‘Macspur’/M7A were the least pre- preferred. ‘Earligold’/EMLA7 and ‘William’s ferred by the panelists (Table 1). Thirty days after Pride’/M7A were not tested due to fruit deterioration storage, ‘Earligold’/EMLA7, ‘Jonafree’/Mark, ‘Mac- (Table 3).

Table 1. Panelists’ ratings of various sensory parameters as influenced by cultivar at harvest, 1999. Parametery Cultivar Appearance Flavor Sweetness Tartness Firmness ‘Earligold’/EMLA7 7.8 ax 6.0 c 5.2 bc 5.6 bcd 4.0 c ‘Jonagold’/EMLA111 8.0 a 8.0 a 7.1 a 7.2 a 7.5 a ‘Improved Golden’/EMLA7 8.0 a 8.0 a 7.5 a 7.3 a 8.0 a ‘Improved Golden’/EMLA111 8.0 a 7.0 b 6.9 a 5.9 bc 7.5 a ‘Scarlet Gala’/EMLA7 8.0 a 7.0 b 5.9 a 6.4 ab 7.5 a ‘Jonafree’/Mark 8.0 a 5.5 d 5.3 bc 5.1 d 5.1 b ‘Macspur’/M7A 6.8 b 5.6 d 5.5 b 5.3 b 5.3 b ‘Royal Gala’/MM111 8.0 a 7.5 ab 6.8 a 6.8 ab 8.0 a ‘William’s Pride’/M7A 8.0 a 6.5 bc 6.7 a 6.0 bc 7.0 a xMeans in columns separated by Duncan’s Multiple Range Test, P < 0.05. Means with the same letter do not differ yParameters were rated on a 10 point scale, where 1 = dislike extremely, and 10 = like extremely

Table 2. Panelists’ ratings of various sensory parameters as influenced by cultivar at 30 days storage time, 1999. Parametery Cultivar Appearance Flavor Sweetness Tartness Firmness ‘Earligold’/EMLA7 5.5 bx 5.9 b 4.4 c 4.5 c 4.5 c ‘Jonagold’/EMLA111 7.5 a 7.6 a 6.7 a 6.8 a 6.8 a ‘Improved Golden’/EMLA7 7.5 a 7.8 a 7.3 a 7.1 a 7.1 a ‘Improved Golden’/EMLA111 6.6 a 6.9 b 6.8 a 5.9 bc 6.9 a ‘Scarlet Gala’/EMLA7 7.5 a 6.7 b 6.5 a 6.0 ab 6.0 ab ‘Jonafree’/Mark 7.0 a 5.0 c 5.0 b 4.4 c 5.4 b ‘Macspur’/M7A 5.5 a 5.2 c 5.0 b 4.8 cd 5.0 b ‘Royal Gala’/MM111 7.0 a 7.0 ab 6.7 a 6.2 ab 7.2 a ‘William’s Pride’/M7A 6.6 a 5.9 b 5.4 b 4.8 c 4.8 c xMeans in columns separated by Duncan’s Multiple Range Test, P < 0.05. Means with the same letter do not differ yParameters were rated on a ten point scale, where 1 = dislike extremely, and 10 = like extremely

July 2005 Vol 50, No. 3 179 Table 3. Panelists’ ratings of various sensory parameters as influenced by cultivars at 60 days storage time, 1999. Parametery Cultivar Appearance Flavor Sweetness Tartness Firmness ‘Earlgold’/EMLA7 —z — — — — ‘Jonagold’/EMLA111 7.1 ax 6.8 a 6.3 a 5.8 b 6.8 a ‘Improved Golden’/EMLA7 6.6 a 7.2 a 6.8 a 6.9 a 7.3 a ‘Improved Golden’/EMLA111 6.6 a 6.6 a 5.4 b 5.9 b 7.0 a ‘Scarlet Gala’/EMLA7 6.9 a 6.5 a 6.1 a 5.8 b 6.6 a ‘Jonafree’/Mark 5.3 b 4.5 b 6.5 a 4.0 c 4.5 b ‘Macspur’/M7A 5.5 b 5.1 b 4.4 c 4.5 b 4.8 b ‘Royal Gala’/MM111 7.3 a 6.7 a 6.5 a 6.2 ab 6.8 a ‘Williams Pride’/M7A —z — — — — xMeans in columns separated by Duncan’s Mulitple Range Test, P < 0.05. Means with the same letter do not differ yParameters were rated on a 10 point scale, where1 = dislike extremely, and 10 = like extremely zNo data presented due to fruit deterioration

Table 4. Maturity indices of apple cultivars measured at harvest time, 1999. Cultivar Diameter (cm) Length (cm) SSC (°Brix) pH Firm (N) ‘Earligold’/EMLA7 67.8 b 59.7 bc 12.8 c 3.63 bc 84.8 d ‘Jonagold’/EMLA111 75.5 a 63.3 a 13.9 ab 3.6 c 113.7 c ‘Improved Golden’/EMLA7 67.4 b 62.5 ab 14.1 a 3.7 b 132.9 bc ‘Improved Golden’/EMLA111 63.3 b 56 .8 c 14.1 a 3.7 b 145.5 b ‘Scarlet Gala’/EMLA7 64.6 b 58.2 c 13.9 ab 3.75 b 151.2 ab ‘Jonafree’/Mark 56.3 c 45.7 d 14.4 a 3.45 d 156.0 a ‘Macspur’/MM111 57.5 c 46.2 d 13.4 b 3.4 d 134.3 bc ‘Royal Gala’/MM111 65.6 b 59.1 c 13.5 b 3.86 a 133.5 bc ‘William’s Pride’/M7A 74.1 a 60.3 abc 13.9 ab 3.84 ab 98.5 d Means separated (by letters) in columns by Duncan’s multiple range test, P > 0.05

At harvest, consumer preference based on free’/Mark which was least preferred. Tartness tartness showed that ‘Jonagold’/EMLA111, ‘Im- ratings for ‘Earligold’/EMLA7 and ‘William’s proved Golden’/EMLA7, ‘Scarlet Gala’/EMLA7, Pride’/M7A decreased considerably in storage for 60 and ‘Royal Gala’/MM111 had the highest rating, days and samples were not evaluated due to fruit while ‘Jonafree’/Mark had the lowest rating or least deterioration. preferred (Table 1). Thirty days after storage, prefer- At harvest, consumer preference based on firm- ence based on tartness showed that ‘Jona- ness showed that ‘Jonagold’/EMLA111, ‘Improved gold’/EMLA111, ‘Improved Golden’/EMLA7 Golden’/EMLA7, ‘Improved Golden’/EMLA 111, ‘Scarlet Gala’/EMLA7 and ‘Royal Gala’/MM111 ‘Scarlet Gala’/EMLA7, ‘Royal Gala’/MM111 and did not differ. The remaining cultivars were least ‘William’s Pride’/M7A did not differ and were preferred (Table 2). Sixty days after storage, ‘Im- preferred compared to ‘Jonafree’/Mark and ‘Mac- proved Golden’/EMLA7 were most preferred, spur’/M7A. ‘Earligold’/EMLA7 was the least pre- followed by the remaining cultivars, except ‘Jona- ferred (Table 3). Thirty days after storage, ‘Jona-

180 Journal of the Mississippi Academy of Sciences gold’/EMLA111, ‘Improved Golden’/EMLA7, ‘Im- ered) had the highest SSC, juice pH, and firmness proved Golden’/EMLA111, ‘Scarlet Gala’/EMLA7 values (Table 4). In addition, fruit deterioration of and ‘Royal Gala’/MM111 were most preferred and ‘Earligold’/EMLA7 and ‘William’s Pride’/M7A two did not differ, followed by ‘Jonafree’/Mark and months after harvest, coincided with the lowest SSC, ‘Macspur’/M7A. ‘Earligold’/EMLA7 was least juice pH, and firmness values of these cultivars preferred. Sixty days after storage, all cultivars were which reflected loss of fruit quality due to senes- equally rated, except ‘Jonafree’/Mark and ‘Mac- cence. spur’/M7A which were least preferred. ‘Earli- gold’/EMLA7 and ‘William’s Pride’/M7A were not CONCLUSION tested due to fruit deterioration. In Canada, it is assumed that consumers find apples with a firmness In general, this study identified ‘Royal of less than 44.5 N too soft (Prange et al., 1993). A Gala’/MM111, ‘Jonagold’/EMLA111, ‘Improved sensory evaluation of ‘McIntosh’ in New York Golden/EMLA7, ‘Improved Golden’/EMLA111, and found crispiness to be directly related to firmness. ‘Scarlet Gala’/EMLA7, as the most preferred culti- Apples with pressure test values of 85.5 N were vars both at harvest and after storage. ‘Jona- rated as “crisp, neither too hard nor too soft” by free’/Mark and ‘Macspur’/M7A were the least consumers, while those with firmness of 31.5 to 36 acceptable cultivars for fresh fruit consumption. N were rated too soft (Lin and King, 1978). ‘Jonafree’ and ‘Macspur’ are progenies of cultivars In general, changes in apple preference with time traditionally used for baking, therefore, such findings in storage differed among the cultivar/rootstock are not surprising. Early harvested cultivars ‘Earli- combinations tested. Such differences are mainly gold’/EMLA7 and ‘William’s Pride’/M7A were due to differences in the physiological age of fruit at identified as having poor keeping quality in storage harvest and losses in fruit quality with time in and may be more suitable for the immediate fresh storage (Wang, 1999). market or short time storage. The consistency in Relating the sensory evaluation to the analytical panelists’ preference as measured by appearance, data of maturity indices at harvest (Table 4), it was flavor, sweetness, tartness, and firmness indicates found that the highest preference in appearance that such parameters are adequate to determine coincided with greater fruit length (r = 0.77, n = 24) consumer acceptance. and greater fruit diameter (r = 0.70, n = 24). The least preferred cultivar/rootstock combination, ‘Mac- ACKNOWLEDGMENTS spur’/M7A, had small fruit. Cultivars intermediate in fruit size, also, maintained a high preference. Factors Thanks to Dr. Clarence Watson for his time and assistance such as color, fruit shape, and cosmetic appearance with the statistical analysis of data, and to the panelists for their were not included in this study, since appearance support in the sensory evaluations in this study. was based on an overall rating of like or dislike. LITERATURE CITED Such factors must be included considering that most consumers use color as an indicator of ripeness Brumfield, R.G., A.O. Adelaja, and K. Lininger. 1993. Con- while others look for uniformity of fruit (Brumfield sumer tastes, preferences, and behavior in purchasing fresh et al., 1993). tomatoes. J. Amer. Soc. Hort. Sci. 118(3):433–438. Soluble solid content (SSC) and juice pH are Janson, H.F. 1972. Taste evaluation of apples from an Ontario find garden. Fruit Var. J. 26(2):26–33. commonly used to evaluate fruit flavor. In this Johnson, D.S., and U. Ertan. 1983. Interaction of temperature study, fruits that had high SSC and high juice pH and oxygen level on the respiration rate of storage quality were not necessarily rated high in flavor, sweetness, of ‘Idared’ apples. J. Hort. Sci. 58:527–533. and tartness. However, using destructive techniques Kadar, A.A. 1999. Fruit maturity, ripening, and quality relation- such as SSC, pH values, and titratable acidity, ships. Prov. Inst. Symp. on effect of pre- and postharvest factors on storage of fruit. Acta Hort. 485:203–208. insures a minimum of quality acceptability for the Kappel F.M., M. Dever, and M.Bouthillier. 1992. Sensory consumer (Kader, 1999). In comparing firmness at evaluation of “Gala” and “Jonagold” strains. Fruit Var. J. harvest using a penetrometer and firmness by the 46(1):37–43. panelists, it was evident that the more acceptable Liu, F.W., and M.M. King. 1978. Common evaluation of cultivars by the panelists were firmer at harvest. ‘McIntosh’ apple firmness. HortScience 13:162–163. Two months after harvest, the most preferred culti- vars in the sensory test (all the parameters consid-

July 2005 Vol 50, No. 3 181 Moore, J.N., and J.R. Ballington Jr. 1990. Genetic resources of nance, and quality evaluation. Acta Hort. 485:389–392. temperate fruit and nut crops I, International Society for Watada, A.E., J.A. Abbott, and R.E. Hardenburg. 1980. Horticultural Science. Wageningen, The Netherlands. 980 Sensory characteristics of apple fruit. J. Amer. Soc. Hort. pp. Sci. 105(3):371–375. Patte, H. (ed). 1985. Evaluation of quality of fruit and vegeta- Watkins, C.B., P.L. Brookfiel, and F.R. Harker. 1993. Develop- bles. AVI Publishing Company, Inc. Westport, Conn. 315 ment of maturity indices for the ‘Fuji’ apple cultivar in pp. relation to watercore incidence. Acta Hort. 326:267– 275. Plotto, A., A.N. Azarenko, J.P. Mattheis, and M.R. McDaniel. Westwood, M.N. 1993. Temperate-Zone Pomology 3rd ed. 1995. ‘Gala’, ‘Braeburn’, and ‘Fuji’ apples: maturity Timber Press. Portland, Oregon. 523 pp. indices and quality after storage. Fruit Var. J. 49:133–142. Williams, A.A., and S.P. Langron. 1983. Influence of different Plotto, A., A.N. Azarenko, M.C. McDaniel, P.W. Crockett, and controlled atmospheres and poststorage temperatures on J.P. Mattheis. 1997. Eating quality of ‘Gala’ and ‘Fuji’ the acceptability of Cox’s Orange Pippin and Suntan apples from multiple harvests and storage durations. apples. J. Sci. Food. Agr. 34:1375–1382. HortScience 32(5):903–908. Wills, R.B.H., T.H. Lee, D. Graham, W.B. McGlasson, and Prange, R.K., M.Meheriuk, E.C. Lougheed, and P.D. Lidster. E.G. Hall. 1998. Postharvest. An introduction to the phys- 1993. Harvest and storage. Pages 64–69 in C.G. Embree. iology and handling of fruits and vegetables. New South Producing apples in eastern and central Canada. Agricul- Wales University Press. The AVI Publishing Company ture Canada, Publication 1899/E. Inc., Westport, Conn. 262 pp. Wang, C.Y. 1999. Postharvest quality decline, quality mainte-

182 Journal of the Mississippi Academy of Sciences Modeling Extinction: Density-Dependent Changes in the Variance of Population Growth Rates

Thomas E. Heering Jr. and David H. Reed1 University of Mississippi, Department of Biology, University, MS, 38677-1848

The ability to accurately predict the likelihood of extinction in endangered populations of is crucial to many concerns in conservation biology. A number of parameters are generally believed to significantly affect a population’s probability of becoming extinct over a given time span. The relationship between the per capita growth rate of a population and its density (i.e., density dependence) is one such parameter. However, the extent to which density dependence influences population dynamics, the usual shape(s) of the density-dependent function, and the impact of density dependence on population persistence, remain controversial. Here we analyze empirical data from 74 populations (40 species) and find evidence for the ubiquity of density dependent population growth. More importantly, using stochastic population models, we find that density-specific changes in the variance of population growth rates have a larger effect on median time until extinction than do changes in the mean population growth rate. Previous studies have focused primarily on density- dependent changes in the mean growth rate. We demonstrate that density-dependent changes in both the mean and the variance of population growth rates can greatly affect the median time to extinction predicted from stochastic population models.

Population growth cannot continue indefinitely function and its interaction with stochastic factors in the face of finite resources. As competition for and life history (Lande et al., 2002; Schoener et al., resources increases at higher population densities, 2003). For example, the inability of individuals to the rate of population growth should slow down and find a mate or engage in group defense at very low eventually stop. Density-dependent population densities might create an absorbing boundary that growth is defined as the dependence of the per capita increases the probability of extinction. population growth rate on past population densities Count-based population viability analyses are (Murdoch and Walde, 1989). Ecologists have de- often used to estimate population persistence, be- bated the generality and importance of density- cause these types of data are the ones most often dependent factors (e.g., malnutrition, disease epi- available to conservation biologists (Morris and demics) to population dynamics for 70 years. Doak, 2002). Given a starting population size (N0), Recent advances in statistical techniques and an the probability of extinction might simply be deter- increase in the number of long-term ecological mined by the mean (:) and the variance (F2) of the studies has led to a growing consensus that density- distribution of population growth rates (see Dennis et dependent reproduction and mortality appears to be al., 1991; Reed and Hobbs, 2004). widespread in natural populations of and However, competing claims have been made invertebrates (e.g., Woiwood and Hanski, 1992; about the utility and reliability of count-based meth- Holyoak, 1993; Wolda and Dennis, 1993; Turchin, ods (Brook, 1999; Ludwig, 1999; Fieberg and Ellner, 1995; Lande et al., 2002), is thought by many to 2000; Meir and Fagan, 2000; Sabo et al., 2004). One greatly influence the probability of population potential problem with count-based models is their persistence, and has long been considered important failure to take into account the dependence of both for population dynamics generally (Ferson et al., the mean and the variance of population growth rates 1989; Hanski, 1990; Burgman et al., 1993; Dennis on population density. This density-dependence, and Taper, 1994; Dennis et al., 1995; Hanski et al., along with correlation patterns among time points in 1996; Lande et al., 2002; Sæther et al., 2002; Henle population size due to the temporal autocorrelation in et al., 2004). However, whether density dependence environmental factors (Pimm and Redfearn 1988; increases or decreases the probability of extinction Inchausti and Halley 2001; Reed et al., 2003a), depends on the exact shape of the density dependent complicates the seemingly simple relationship

1Author for correspondence. Department of Biology, P. O. Box 1848; [email protected]

July 2005 Vol 50, No. 3 183 between time until extinction and the mean and be in a continual growth phase (i.e., the median variance of population growth rates. population growth rate was approximately zero). At the heart of this debate is the question of how Extensive analysis of the data sets in the GPDD complex population viability analysis models need (Inchausti and Halley, 2001; Reed et al., 2003a,b; to be in order to accurately convey population Reed, 2004; Reed and Hobbs, 2004) have shown dynamics. Modeling population dynamics is crucial these are the data sets appropriate for answering the to determining minimum viable population sizes and questions being addressed in this paper. ranking conservation priorities. Whether estimates Population growth rates (r), for each time step, of the density-dependence of population growth were calculated using the following formula: rates are necessary for accurate and unbiased esti- mates of extinction risk is a question of current rt = loge (Nt / Nt-1) (1), concern among conservation biologists (e.g., Henle et al., 2004; Sabo et al., 2004). where Nt is population size at time t. Positive statisti- Despite the importance of the variance in popu- cal outliers (especially if they did not seem possible lation growth rates, and variance in demographic given the species life history) were rare and were parameters generally, to all forms of population removed when detected. However, the distribution of viability analysis, we are not aware of any published growth rates was normally or approximately nor- empirical results addressing changes in the variance mally distributed in all but four cases. Once a distri- among population growth rates with changes in bution of r-values had been calculated the distribu- density. However, variance in growth rates has been tion was assayed for evidence of density dependence. included in a very general way in both diffusion Thus, Nt-1 was regressed against rt. This regression approximations and continuous time Markov chain was allowed to be a first, second, or third order models of population dynamics (e.g., Mangel and polynomial. We used corrected Akaike Information Tier 1993; Wilcox and Elderd, 2003). Here, we Criterion statistics (Burnham and Anderson, 2002) examine changes in the variance of population and model averaging to identify the best fit model. growth rates, as well as changes in the mean growth No time lags were allowed in the density depend- rate, with density. Further, we demonstrate what ence. Solving for the carrying capacity (K) was effect these changes have on the probability of carried out by setting rt equal to zero and solving for extinction, as compared to models that disregard N. Thus, the carrying capacity is defined as the density-dependence in population growth rates population size (density) where the mean population entirely and to those that only consider changes in growth rate is expected to be zero. the mean population growth rate. Once values for K were calculated, means and standard deviations for population growth rates were MATERIALS AND METHODS determined for each population at three different density categories: when N $ K (high density), when Stochastic, discrete time, discrete state popula- 0.5K # N < K (intermediate density), and when N < tion models were built from empirical data on the 0.5K (low density). Coefficients of variation in the distribution of population growth rates, at different population growth rate (SDr / r) were also calculated, densities, estimated from census data obtained from for each population, at each of the three density the Global Population Dynamics Database categories. (GPDD)(NERC 1999) for 74 populations (40 spe- In order to examine the effects of both the mean cies). The populations used to build the models were and the standard deviation of the population growth chosen based on the following criteria: (1) The rate changing with density, we developed a set of quality of the data as determined by the GPDD and stochastic discrete time, discrete state, r-models the length of the census period assayed. The reliabil- calibrated from empirical data. The basic format of ity of the census data is scored by the GPDD from the model is as follows. First, an initial population one to five (with five being the most reliable). Data size (N0) was set equal to the initial population size sets were selected for inclusion in the study when from the actual time series. N0 was compared to K the reliability rating multiplied by the census period and a population growth rate (r) randomly chosen $ 50. (2) The population was judged to be in a stable from a normal distribution with a mean determined equilibrium. Thus, the population was not in contin- from the regression function and a standard deviation uous decline or so far below carrying capacity as to estimated from the actual distribution of growth rates

184 Journal of the Mississippi Academy of Sciences from the time series. The growth rate is then used to ized beta values were calculated to rank the signifi- determine population size (Nt) in the next time step. cant factors effects on median time to extinction The process is repeated for each time step, with the (Neter et al., 1996). distribution of possible randomly selected r values Sensitivity analysis was performed on the for each time step being determined by the ratio of density-specific standard deviation in population N:K and the regression function. Each model was growth rates by increasing the standard deviation by run for at least 1,000 simulations. 10% increments (up to a maximum increase of 50%) In order to estimate the median time to extinc- for each density category separately and estimating tion, we used at least 1,000 simulations of each the slope of the best fit linear line using the standard population. Each simulation had identical starting deviation in the population growth rate as the inde- points and the was allowed to run stochastically for pendent variable and median time to extinction as the a fixed number of time steps (years). The proportion dependent variable (Morris and Doak, 2002). The of replicate populations going extinct within the slopes were averaged across all 74 populations and determined time frame was recorded. The number of then compared for the three different density catego- time steps was then varied until several estimates of ries. extinction probability above and below 50% were To examine the impact of ignoring density generated. From this data, linear regression was used dependence altogether or allowing only the mean to estimate the number of time steps sufficient for growth rate to change, relative to the full model 50% of the populations to go extinct (median time to where the mean and variance in growth rates were extinction). The median time to extinction is cer- allowed to change with density, we built additional tainly not realistic, as no age-structure, explicit models for 25 randomly chosen species. Thus, two demographic stochasticity, or genetic stochasticity additional models were constructed with either no was included in the models. However, it still has density dependence or density dependence where much heuristic value as concerns the effects of only the mean population growth rate changes with density dependence. density. Median time to extinction was estimated for We used corrected Akaike Information Criterion models with no density dependence (NDD) and for statistics and model averaging (Burnham and Ander- models where only the mean population growth rate son, 2002) to identify what parameters are important was allowed to change with changing density (SDD). with respect to the median time to extinction. Back- For NDD, the mean population growth rate and the wards stepwise multiple regression was used to variance among growth rates was the same regardless estimate the parameter coefficients from the consen- of density and the values were estimated from the sus model. Thirteen variables were initially tested entire census period. For SDD, the variance among across different model combinations: Initial popula- growth rates was the same regardless of density, but tion size, the carrying capacity (K), the median the mean growth rate changed with density according growth rate for the entire census period (rmed), the to the regression function. mean growth rate at high densities, the mean growth We also used an analysis of covariance (using rate at intermediate densities, and the mean growth carrying capacity as the covariate) to examine whe- rate at low densities, the maximum value of r during ther there were broad phylogenetic (Class, Order, the census period (rmax), the standard deviation in Family) or environmental (biogeographic region, the population growth rate at high densities, the global latitude) effects on the mean time to extinc- standard deviation in the growth rate at intermediate tion. densities, the standard deviation in the population growth rate at low densities, the coefficient of RESULTS variation in the growth rate at high densities, the coefficient of variation in the growth rate at interme- A meta-analysis of 74 populations was conducted diate densities, and the coefficient of variation in the with respect to how the mean, standard deviation, population growth rate at low densities. Because and coefficient of variation in population growth standard deviations and coefficients of variation are rates changes with changes in population density not independent measures, model selection was (Table 1). The mean population growth rate is signif- based on models that included either the standard icantly different at all three density categories and deviation or the coefficient of variation, not both. decreases with increasing density across all the 74 Once the important factors were identified, standard- populations. Likewise, the standard deviation among

July 2005 Vol 50, No. 3 185 ance among population growth Table 1. Means and standard errors are presented for three different rates as well. parameters for three different density ranges. We performed model selec- tion using the information-the- Parameter Mean ± SE F P oretic approach of Burnham and Anderson (2002), using CVr (high) 0.331 ± 0.045 30.57 < 0.0001 independent combinations of CVr (intermediate) 0.683 ± 0.045 13 different model parameters CV (low) 0.205 ± 0.031 (Table 3). The consensus r model containing five factors r (high) -0.155 ± 0.021 (all significant using multiple r (intermediate) 0.082 ± 0.013 123.15 < 0.0001 regression) are listed and ranked according to their stan- r (low) 0.356 ± 0.035 dardized beta values. The sig- nificant factors, from greatest SD (high) 0.251 ± 0.016 r effect to least effect are: SD 11.29 < 0.0001 r SDr (intermediate) 0.355 ± 0.025 (intermediate densities) (F = 23.48, P < 0.0001, b = -0.431), SDr (low) 0.515 ± 0.047 K (F = 31.69, P < 0.0001, b =

0.398), SDr (low densities) (F growth rates at a given density significantly de- = 9.08, P < 0.005, b = -0.242), r (intermediate densi- creases with increasing density. The coefficient of ties) (F = 8.13, P < 0.005, b = 0.196), and r (low variation is also highly significantly different for densities) (F = 7.52, P < 0.01, b = 0.186). The overall each density category, but the maximum coefficient regression explains 70.0% of the variation in median of variation is reached at intermediate densities. extinction times (adjusted R2= 0.700). Thus, density-dependent effects on population In the Introduction it was suggested that the growth rate were consistent and highly significant shape of the density-dependent function for growth despite the data being noisy (e.g., differences in rates was important to whether it decreased the species biology and generation length, differences in probability of extinction (primarily believed to be the quality of the data). Further, these density- true) or increased the probability of extinction as dependent changes were not just to the mean popula- might be true with strong Allee effects. We illustrate tion growth, but also to two measures of variation in the five general forms of density dependence in growth rates. population growth rate found in this study and give It is important to be able to separate purely their relative frequencies (Figure 1). A linear model demographic causes of variation in growth rates was found to be the best fit function for 45 of 74 from those brought about by the effects of density. populations. However, the statistical power to detect Table 2 provides a comparison between how the nonlinearities in individual data sets was often low. standard deviation among population growth rates, Thus, the lack of evidence for general nonlinearity in for a given density category, changes in large (K > the relationship between density and per capita 200) versus small (K # 200) populations. Standard growth rates in these data sets should not be con- deviations significantly increase with decreasing strued as suggesting that such nonlinearities do not density in both large and small populations, thus exist. A 3rd degree polynomial with population there is an effect of den- sity that is independent of Table 2. Comparison of the mean (with standard error) standard deviation demographic stochas- in population growth rates across the 74 populations, for three different ticity. However, the stan- density categories, divided as to whether the carrying capacity was less dard deviation is consis- than or greater than 200 individuals. tently larger in smaller $ $ populations, indicating SDrr (N > K) SD (K N 0.5K) SDr (N < 0.5K) that demographic stochas- K $ 200 0.18 ± 0.02 0.30 ± 0.04 0.35 ± 0.04 ticity plays a significant role in the amount of vari- K < 200 0.28 ± 0.02 0.40 ± 0.03 0.58 ± 0.06

186 Journal of the Mississippi Academy of Sciences growth rates increasing at an increasing rate at very density or population growth rates where the mean low densities, and decreasing at an increasing rate at and variance were allowed to change with respect to very high densities, was the best fit model in 19 of density. The median extinction times for the three 74 populations. A 2nd degree polynomial where models are significantly different from each other (P population growth rates increase at an increasing < 0.01), with the models that allow both the mean rate at low densities was the best fit in 6 of 74 and variance in population growth rates to change populations. A 2nd degree polynomial where popula- having the longest median times to extinction and tion growth rates decrease at an increasing rate at those having no density dependence the shortest. both high and low densities was the best fit in 3 of This paper uses the meta-analysis technique to 74 populations. A 2nd degree polynomial where the look for factors useful in predicting extinction risk population growth rate decreases at an increasing for populations of vertebrates. These techniques are rate at high densities was the best fit in only 1 of 74 especially important when individual studies cannot populations. be generalized and lack statistical power. However, An important question in conservation biology is it is often useful to test for any patterns based on the how complex population viability models need to be ecology, life history, or evolutionary histories of the in order to predict the risk of extinction accurately organisms. We find no differences in median time to and without bias. Table 4 shows the results of an extinction based on phylogeny (Class, Order, Family) analysis of variance comparing median time to or environmental classification (biogeographic extinction, for 25 randomly chosen population region, global latitude). None of these factors were models, using three different modeling approaches significant once the effects of carrying capacity were with the models each developed from the same set of accounted for. These results are congruent with observed data. The models contained either no other studies that have looked for these types of density dependence, density dependent population effects (Gaillard et al., 2000; Inchausti and Halley, growth where the mean growth rate changes with 2001; Reed et al., 2003a; Reed and Hobbs, 2004).

Table 3. The results of multiple regression Table 4. Results from an analysis of variance analysis examining 13 factors suspected of being (randomized block design and Tukey’s HSD important in determining median time to ex- test), comparing median extinction times (in tinction in 74 population viability models cre- years) for 25 species. Models were built with ated from census data on natural populations of either no density dependence (NDD), density animals. The significant parameters from each dependence where only the mean population model are listed in order of importance as growth rate changes with density (DDM), and determined by their standardized beta values density dependence where both the mean and (adjusted R2 = 0.700, p < 0.0001). standard deviation of the population growth rate were allowed to change with changes in density Parameter Probability Std Beta (DDMS). Each of the three model assumptions

SDr (medium) < 0.0001 -0.431 leads to significantly different median times to extinction (MT ) (F = 18.95, p < 0.001). log K < 0.0001 0.398 E Model MTE SDr (low) 0.0036 -0.242 DDM 476.4 r (medium) 0.0036 0.196 DDMS 913.1 r (low) 0.0108 0.186 NDD 126.2

July 2005 Vol 50, No. 3 187 A B

C D

E Figure 1. Five general forms of density dependence in population growth rate. (a) Linear was the best fit function for 45 of 74 populations. Note the decreasing variation in population growth rates with increasing density. (b) A 3rd degree polynomial with population growth rates increasing at an increasing rate at low densities, and decreasing at an increasing rate at high densities, was the best fit in 19 of 74 populations. (c) a 2nd degree polynomial where population growth rate increases at an increasing rate at low densities was the best fit in 6 of 74 populations. Note the greater variance in growth rates at lower densities. (d) a 2nd degree polynomial demonstrating Allee effects was the best fit in 3 of 74 populations. (e) a 2nd degree polynomial where population growth rate decreases at an increasing rate at high densities was the best fit in only 1 of 74 populations. Note the greater variance in growth rates at lower densities. 188 Journal of the Mississippi Academy of Sciences DISCUSSION increase in the variance among growth rates at lower densities, despite their being so large (even at their Ubiquity of density dependence. The role of lowest observed densities) as to make demographic density-dependent mortality and birth rates, in stochasticity almost nonexistent. Thus, there seems impacting population dynamics, has been a source of to be a component of the variance among population controversy for at least 70 years (Turchin, 1995). growth rates that is driven by population densities However, there seems to be a growing consensus and not just population size. This suggests, as one that most populations (at least at times) are regulated possibility, that widespread Allee effects may impact through density-dependent mechanisms in conjunc- not just the mean but the variance in population tion with stochastic factors (e.g., Turchin, 1995; growth rates. Sæther et al., 2002; Reed et al., 2003a). There exist Parameters affecting median time to extinction. far more sophisticated statistical tests for detecting Five factors were identified as significantly affecting density-dependence than the one we use here (e.g., median time to extinction in our models, in rank Dennis and Taper, 1994) and the question of what order of their standardized beta values they are: the role density dependence plays in population dynam- standard deviation among population growth rates at ics will not be answered definitively by this study. intermediate densities, the carrying capacity, the However, it is worth noting that the simple regres- standard deviation among population growth rates at sion techniques we used, with an assumption of no low densities, the mean population growth rate at time lags and that errors are additive, was significant intermediate densities, and the mean population in 58 of 74 populations examined (78.4%). This is growth rate at low densities. true despite the median time series being only 19 It is certainly not surprising to see that carrying years and the fact that statistical power is reduced if capacity is a major factor affecting median time to these assumptions are violated (Turchin, 1995). The extinction. Models of population viability are usually mean percent of the variance in the per capita sensitive to changes in carrying capacity and there is population growth rates explained by population size plenty of empirical data linking larger population in the previous time step was 29% (SE ± 1.9%) size to a greater probability of population persistence across all 74 populations. Mean population growth (see review in Reed et al., 2003a; O’Grady et al., rates clearly, consistently, and significantly declined 2004). The univariate regression gives the following as densities increased. Thus, though not a robust test formula for time to extinction at a given carrying of the presence of density dependence, this study capacity: suggests that density dependence is widespread across populations (Appendix I & II). log10 MTE = 0.9135 + 0.5776 (log10 K) (2), Density dependent changes in the variance of population growth rates. In addition to changes in where MTE is the median time to extinction in years the mean per capita population growth rate, changes and K is the carrying capacity. However, this model in the variance among growth rates at different undoubtedly overestimates median time to extinction. densities were detected. Lower densities led to larger In fact, these stochastic r models predict median standard deviations among population growth rates. extinction times that are more than five times as long This is expected for demographic reasons alone, (K = 10), three times as long (K = 100), or 1.5 times because smaller populations are more variable as long (K = 10,000) as models that incorporate far (Taylor et al., 1980; Reed and Hobbs, 2004). Indeed, greater complexity (Reed et al., 2004). the data confirm that smaller populations tend to The standard deviation in the population growth have greater standard deviations at all three density rate is more important than the mean of the popula- categories (significantly higher at the highest and tion growth rate, in determining median time to lowest densities) than do larger populations. How- extinction. Thus, models that are deterministic or do ever, there are significant changes in the standard not carefully consider estimates of the variation in deviation among population growth rates with population growth rates, or demographic parameters changes in density for the large populations and the generally, will not be able to provide accurate infor- same pattern of increasing variance among growth mation on the probability of extinction. rates with decreasing densities can be seen. Anec- The median time to extinction was most affected dotally, even populations of tens of thousands of by both the standard deviation in population growth individuals still often showed the characteristic rates and the mean population growth rate at interme-

July 2005 Vol 50, No. 3 189 diate densities. This is contrary to intuition, as it (Pimm and Redfearn, 1988; Reed et al., 2003a), the might be expected that populations would be most lack of inclusion of rare catastrophic events that vulnerable to extinction when they are at their greatly impact population persistence (Reed et al., lowest densities (smallest size). It is possible that 2003b), and density dependent changes in the mean this result is simply due to there being so much more and variance of population growth rates. variation among models for the parameters at this density. With this hypothesis in mind, we conducted ACKNOWLEDGMENTS sensitivity analysis on changes in the standard deviation among growth rates for all three densities We thank the Mississippi Space Grant Consortium and the for 30 randomly chosen models. Using multiple National Aeronautics and Space Agency for their financial support for this project. We also thank three anonymous regression, we found that the models were most reviewers for their helpful comments regarding a prior version sensitive to changes in the standard deviation at the of this paper. lowest densities (data not shown) as expected from theory. The differences in sensitivity at low and LITERATURE CITED intermediate densities was small, but statistically significant. Brook, B.W. 1999. Evaluating population viability analysis. Importance of including density-dependent Ph.D. thesis, Macquarie University, Sydney. Burgman, M.A., S. Ferguson, and H.R. Akçakaya.1993. Risk changes in the mean and variance. We are not the assessment in conservation biology. Chapman & Hall, first to suggest that density dependence is an impor- London. tant component to include in population viability Burnham, K.P., and D.R. Anderson. (2002) Model selection and models (see Introduction). In our simple count-based multimodel inference: A practical information-theoretic nd population viability analysis for 25 species, the approach, 2 edition. Springer, New York. Dennis, B., R.A. Desharnais, and J.M. Scott. 1991. Estimation median time to extinction without density depend- of growth and extinction parameters for endangered ence was less than 30% of the median time to species. Ecol. Monogr. 61:115–143. extinction when density dependence was included. Dennis, B., P.L. Munholland, J.M. Cushing, and R.F. Constan- Thus, the models without density dependence were tino. 1995. Nonlinear demographic dynamics: Mathemati- cal models, statistical methods, and biological experiments. not just pessimistic, but they were extremely pessi- Ecol. Monogr. 65:261–281. mistic relative to the models with density depend- Dennis, B. and B. Taper. 1994. Density dependence in time ence. This suggests that density dependence gener- series observations of natural populations: Estimation and ally “puts the brakes on” declining populations by testing. Ecol. Monogr. 64:205–224. creating reflecting rather than absorbing points at Ferson, S., L. Ginzburg, and A.A. Silvers. 1989. Extreme event risk analysis for age-structured populations. Ecol. Model. low densities. 47:175–187. Including changes in the variance among popula- Fieberg, J., and S.P. Ellner. 2000. When is it meaningful to tion growth rates, in addition to changes in the mean estimate an extinction probability? Ecology 81:2040–2047. growth rate, nearly doubles persistence time over the Gaillard, J.-M., M. Festa-Bianchet, N.G. Yoccoz, A. Loison, and C. Tiogo. 2000. Temporal variation in fitness compo- case where only the mean growth rate is allowed to nents and population dynamics of large herbivores. Ann. change with density. This seems counterintuitive at Rev. Ecol. System. 31:367–393. first, given that the variance among growth rates Hanski, I. 1990. Density dependence, regulation and variability increases as population sizes decline and this is in populations. Philos. R. Soc. London, Ser. B precisely when populations are most vulnerable. 330:141–150. Hanski, I., P. Foley, and M. Hassell. 1996. Random walks in a However, the reason for this is that the density- metapopulation: How much density dependence is neces- specific variance, even at low densities, is less than sary for long-term persistence? J. Anim. Ecol. 65:274–282. the variance among population growth rates for the Henle, K., S. Sarre, and K. Wiegand. 2004. The role of density entire census period. Thus, extreme caution must be regulation in extinction processes and population viability analysis. Biodiv. Conserv. 13:9–52. used in building count-based PVAs from census data Holyoak, M. 1993. The frequency of detection of density even in equilibrium populations. Simply computing dependence in insect orders. Ecol. Entom. 18:339–347. the mean and variance of the growth rates over a Inchausti, P. and J. Halley. 2001. Investigating long-term given number of time steps is not likely to produce ecological variability using the global population dynamics the type of dynamics and, therefore, extinction database. Science 293:655–657. Lande, R., S. Engen, B.–E. Sæther, F. Filli, E. Matthysen, and probabilities that exist in natural populations. The H. Weimerskirch. 2002. Estimating density dependence reasons why this is true include the autocorrelation from population time series using demographic theory and structure in environmental variation through time life-history data. Am. Nat. 159:321–337.

190 Journal of the Mississippi Academy of Sciences Ludwig, D. 1999. Is it meaningful to estimate a probability of 2003b. The frequency and severity of catastrophic die-offs extinction? Ecology 80:298–310. in vertebrates. Anim. Conserv. 6:109–114. Mangel, M. and C. Tier. 1993. A simple direct method for Reed, D.H., J.J. O’Grady, B.W. Brook, J.D. Ballou, and R. finding persistence times of populations and application to Frankham. 2003a. Estimates of minimum viable population conservation problems. Proc. Natl. Acad. Sci. sizes vertebrates and factors influencing those estimates. 90:1083–1086. Biol. Conserv. 113:23–34. Meir, E. and W.F. Fagan. 2000. Will observation error and Reed, D.H., J.J. O’Grady, B.W. Brook, J.D. Ballou, and R. biases ruin the use of simple extinction models? Conserv. Frankham. 2004. Reality checks, probability statements, Biol. 14:148–154. and the case of the Bali tiger. Conserv. Biol. 18:1179. Morris, W.F., and D.F. Doak. 2002. Quantitative conservation Sabo, J.L., E.E. Holmes, and P. Kareiva. 2004. Efficacy of biology: Theory and practice of population viability simple viability models in ecological risk assessment: analysis. Sinauer Associates Inc., Sunderland. Does density dependence matter? Ecology 85:328–341. Murdoch, W.W. and S.J. Walde. 1989. Analysis of insect Sæther, B.–E., S. Engen, R. Lande, M. Visser, and C. Both. population dynamics. Pages 113–140 in P.J. Grubb & J.B. 2002. Density dependence and stochastic variation in a Whittaker, eds. Towards a more exact ecology. Blackwell, newly established population of a small songbird. Oikos Oxford. 99:331–337. NERC (1999) NERC & Centre for Population Biology, Impe- Schoener, T.W., J. Clobert, S. Legendre, and D.A. Spiller. rial College. The Global Population Dynamics Database. 2003. Life-history models of extinction: A test with island http://www.Sw.ic.ac.uk/cpb/gpdd.html. spiders. Am. Nat. 162:558–573. Neter, J., M.H. Kutner, and W. Wasserman. 1996. Applied Taylor, L.R., Woiwood, I.P., and Perry, J.N. 1980. Variance Linear Regression Models, 3rd edition. Irwin / McGraw- and the large scale spatial stability of aphids, moths, and Hill, publisher. birds. Journal of Animal Ecology 49:831–854. O’Grady, J.J., D.H. Reed, B.W. Brook, R. Frankham. 2004. Turchin, P. 1995. Population regulation: Old arguments and What are the best correlates of extinction risk? Biol. new synthesis. Pages 19–40 in N. Cappucino and P.W. Conserv. 118:513–520. Price, eds. Population dynamics. Academic Press, New Pimm, S.L., and A. Redfearn. 1988. The variability of popula- York. tion densities. Nature 334:613–615. Wilcox, C., and B. Elderd. 2003. The effect of density-depend- Reed, D.H. 2004. Extinction risk in fragmented habitats. Anim. ent catastrophes on population persistence time. J. Appl. Conserv. 7:181–191. Ecol. 40:859–871. Reed, D.H. and G.R. Hobbs. 2004. The relationship between Woiwood, I.P., and I. Hanski. 1992. Patterns of density depend- population size and temporal variability in population size. ence in moths and aphids. J. Anim. Ecol. 61:619–629. Anim. Conserv. 7:1–8. Wolda, H., and B. Dennis. 1993. Density dependence tests, are Reed, D.H., J.J. O’Grady, J.D. Ballou, and R. Frankham. they? Oecologia 95:581–591.

July 2005 Vol 50, No. 3 191 Appendix I: Class, order, and biogeographic zone for each of the 40 species modeled. Species Class Order Biogeographic Zone Vanellus vanellus Aves Charadriiformes Palaearctic Rissa tridactyla Aves Charadriiformes Nearctic Zenaida macroura Aves Columbiformes Nearctic Accipiter nisus Aves Falconiformes Palaearctic Falco rusticolus Aves Falconiformes Nearctic Alauda arvensis Aves Passeriformes Palaearctic Corvus corone Aves Passeriformes Palaearctic Corvus frugilegus Aves Passeriformes Palaearctic Cyanocitta cristata Aves Passeriformes Palaearctic Melospiza melodia Aves Passeriformes Nearctic Spizella pusilla Aves Passeriformes Nearctic Fringella coelebs Aves Passeriformes Palaearctic Fringella montifringella Aves Passeriformes Palaearctic Anthus pratensis Aves Passeriformes Palaearctic Ficedula albicollis Aves Passeriformes Palaearctic Ficedula hypoleuca Aves Passeriformes Palaearctic Parus atricapillus Aves Passeriformes Nearctic Parus bicolor Aves Passeriformes Nearctic Parus caeruleus Aves Passeriformes Palaearctic Parus major Aves Passeriformes Palaearctic Sturnus vulgaris Aves Passeriformes Palaearctic Phylloscopa collybita Aves Passeriformes Palaearctic Phylloscopa trochilus Aves Passeriformes Palaearctic Phalacrocorax aristotellis Aves Pelecaniformes Palaearctic Picoides pubescens Aves Piciformes Nearctic Ovis canadensis Mammalia Artiodactyla Nearctic Tragelaphus strepsiceros Mammalia Artiodactyla Ethiopian Dama dama Mammalia Artiodactyla Palaearctic Canis lupus Mammalia Carnivora Palaearctic Lynx canadensis Mammalia Carnivora Nearctic Enhydra lutris Mammalia Carnivora Nearctic Gulo gulo Mammalia Carnivora Nearctic Martes americana Mammalia Carnivora Nearctic Phoca groenlandica Mammalia Carnivora Nearctic Phoca vitulina Mammalia Carnivora Palaearctic Ursus arctos horribilis Mammalia Carnivora Nearctic Microtus californicus Mammalia Rodentia Nearctic Merlangus merlangius Osteichthyes Gadiformes Palaearctic Perca fluviatalis Osteichthyes Perciformes Palaearctic Esox lucius Osteichthyes Salmoniformes Palaearctic

192 Journal of the Mississippi Academy of Sciences Appendix II: F(x) = shape of density dependent function (see Figure 1), r2 = proportion of variance in

the population growth rate explained by density in the preceding time step, MTE = median time to

extinction, K = carrying capacity, r (Kn) = the mean population growth rate at high, intermediate and

low densities (K12, K , and K3, respectively), and SDr (Kn) = the standard deviation among population

growth rates at high, intermediate, and low densities (K12, K , and K3, respectively). 2 Species F(x) r MTE K r (Kr1) (K2) r (K3) SDr (K1) SDr (K2) SDr (K3) Vanellus vanellus a 0.06 8 13 -0.033 0.078 0.143 0.382 0.496 0.685 Vanellus vanellus a 0.12 33 19 -0.249 0.151 0.339 0.287 0.292 0.659 Rissa tridactyla b 0.39 919 171 -0.040 0.069 0.181 0.120 0.112 0.198 Zenaida macroura d 0.45 33 24 -0.395 0.172 0.619 0.117 0.527 0.516 Zenaida macroura a 0.34 21 110 -0.147 0.146 0.321 0.516 0.415 0.214 Accipiter nisus a 0.08 329 111 -0.055 0.027 0.033 0.077 0.197 0.047 Accipiter nisus a 0.87 710 52 -0.058 0.223 1.035 0.093 0.055 2.309 Falco rusticolus b 0.20 144 82 -0.041 0.007 0.297 0.308 0.323 0.488 Alauda arvensis c 0.48 213 57 -0.043 0.095 0.353 0.124 0.223 0.719 Corvus corone c 0.44 181 17 -0.109 -0.020 0.139 0.160 0.231 0.266 Corvus frugilegus a 0.24 58 355 -0.073 -0.044 0.051 0.075 0.214 0.249 Corvus frugilegus a 0.54 315 126 -0.121 -0.049 0.249 0.174 0.179 0.234 Cyanocitta cristata a 0.24 246 26 -0.090 0.073 0.252 0.206 0.267 0.372 Melospiza melodia a 0.57 26 52 -0.200 0.281 0.568 0.417 0.365 0.096 Spizella pusilla b 0.29 273 65 -0.107 0.031 0.185 0.326 0.332 0.287 Fringella coelebs a 0.09 1581 616 -0.028 0.016 0.047 0.100 0.062 0.147 Fringella montifringilla a 0.19 55 71 -0.109 -0.030 0.365 0.399 0.371 0.463 Anthus pratensis a 0.34 252 88 -0.096 0.017 0.079 0.123 0.142 0.301 Ficedula albicollis a 0.36 9 8 -0.139 0.034 1.206 0.440 0.335 1.706 Ficedula albicollis a 0.03 834 99 -0.131 0.042 0.184 0.184 0.214 0.060 Ficedula hypoleuca a 0.28 10 9 -0.238 0.188 0.495 0.302 0.619 0.746 Ficedula hypoleuca d 0.87 444 82 -0.092 0.184 0.758 0.133 0.269 0.719 Ficedula hypoleuca c 0.23 551 144 -0.046 -0.022 0.141 0.117 0.154 0.146 Parus atricapillus b 0.58 498 114 -0.158 0.340 0.525 0.257 0.387 0.526 Parus bicolor a 0.32 16 8 -0.237 0.067 0.625 0.275 0.458 0.744 Parus bicolor d 0.10 5 12 -0.294 -0.141 0.116 0.131 0.516 0.729 Parus bicolor a 0.37 7 17 -0.210 -0.015 1.333 0.347 0.666 1.282 Parus caeruleus b 0.46 46 46 -0.075 0.113 0.359 0.284 0.479 0.535 Parus caeruleus a 0.55 96 82 -0.193 0.136 0.378 0.203 0.451 0.365 Parus caeruleus a 0.25 745 74 -0.067 0.085 0.436 0.228 0.345 0.302 Parus caeruleus b 0.35 81 89 -0.169 0.168 0.259 0.300 0.333 0.517 Parus caeruleus a 0.24 53 44 -0.122 0.098 0.367 0.193 0.389 0.621 Parus caeruleus b 0.22 151 87 0.037 0.027 0.199 0.399 0.290 0.254 Perca fluviatalis a 0.41 9 18 -0.037 0.031 0.481 0.210 0.211 0.105 Perca fluviatalis a 0.14 41 180 -0.184 0.127 0.261 0.294 0.675 0.145

July 2005 Vol 50, No. 3 193 Appendix II: Cont’d.

2 Species F(x) r MTE K r (Kr1) (K2) r (K3) SDr (K1) SDr (K2) SDr (K3) Parus major a 0.45 131 126 -0.307 0.353 0.709 0.254 0.469 0.387 Parus major a 0.41 293 208 -0.246 0.092 0.366 0.090 0.372 0.295 Parus major a 0.25 103 27 -0.087 -0.065 0.282 0.249 0.158 0.495 Parus major b 0.21 468 94 -0.015 0.038 0.329 0.154 0.230 0.405 Sturnus vulgaris c 0.27 188 54 -0.026 0.103 0.222 0.188 0.395 0.431 Sturnus vulgaris b 0.19 41 62 -0.140 0.241 0.402 0.434 0.500 0.685 Sturnus vulgaris a 0.35 356 61 -0.269 0.032 0.268 0.189 0.233 0.280 Phylloscopa collybita a 0.37 92 12 -0.210 0.091 0.469 0.248 0.288 0.460 Phylloscopa trochilus b 0.45 5 7 -0.441 -0.032 0.500 0.498 0.302 1.389 Phylloscopa trochilus a 0.13 17 9 -0.059 0.038 0.260 0.333 0.461 0.522 Phalacrocorax aristotellis e 0.22 1258 399 -0.019 0.002 0.286 0.046 0.326 0.433 Picoides pubescens a 0.33 54 5 -0.098 -0.052 0.597 0.494 0.269 0.392 Picoides pubescens a 0.16 39 8 -0.080 0.056 0.348 0.273 0.423 0.429 Ovis canadensis b 0.21 1192 185 -0.033 0.039 0.199 0.235 0.273 0.375 Tragelaphus strepsiceros a 0.06 92 6327 -0.086 0.003 0.027 0.097 0.153 0.191 Tragelaphus strepsiceros a 0.46 1387 58502 -0.045 -0.006 0.260 0.147 0.098 0.405 Dama dama a 0.38 1987 970 -0.028 -0.015 0.091 0.088 0.074 0.065 Canis lupus a 0.14 44 399 -0.106 0.246 0.193 0.322 0.526 0.499 Lynx canadensis c 0.20 342 3598 -0.072 0.111 0.355 0.264 0.386 0.594 Lynx canadensis a 0.19 1557 31915 -0.074 0.044 0.462 0.207 0.507 0.805 Lynx canadensis b 0.05 1261 42300 -0.186 0.081 0.235 0.257 0.172 0.353 Enhydra lutris a 0.22 1563 1753 -0.096 0.080 0.072 0.199 0.195 0.170 Gulo gulo b 0.05 417 682 -0.033 0.032 0.056 0.138 0.184 0.308 Gulo gulo b 0.07 1692 799 -0.139 0.041 0.176 0.167 0.151 0.399 Martes americana a 0.26 15 73 -0.136 -0.095 0.625 0.558 0.227 0.873 Martes americana a 0.12 552 44958 -0.117 0.136 0.158 0.264 0.367 0.315 Martes americana a 0.29 779 168 -0.130 0.152 0.279 0.158 0.277 0.202 Phoca groenlandica c 0.42 108 96 -0.230 0.225 0.756 0.157 0.475 0.076 Phoca vitulina b 0.38 2368 1537 -0.001 0.006 0.061 0.067 0.071 0.257 Phoca vitulina b 0.56 843 1208 -0.772 0.058 0.070 0.097 0.104 0.061 Phoca vitulina b 0.27 538 135 -0.089 -0.034 0.002 0.122 0.137 0.075 Ursus arctos horribilis a 0.26 1352 81 -0.052 0.007 0.083 0.189 0.078 0.264 Microtus californicus a 0.11 3 54 -0.530 0.300 1.070 0.396 1.246 2.099 Microtus californicus b 0.14 11 311 -0.044 0.200 0.317 0.214 0.702 0.848 Merlangus merlangius a 0.30 211 1619 -0.432 0.497 0.324 0.237 0.613 0.222 Esox lucius a 0.33 349 1967 -0.136 0.046 0.046 0.209 0.244 0.154 Esox lucius a 0.19 221 2895 -0.044 -0.038 0.336 0.216 0.287 0.643

194 Journal of the Mississippi Academy of Sciences New Records for the Phlebotomine Sand Fly Lutzomyia shannoni (Dyar) (Diptera: Psychodidae) in Mississippi

Jerome Goddard1 and Chad P. McHugh

Mississippi Department of Health, Jackson, MS 39215, and Air Force Institute for Operational Health, Brooks City-Base, TX 78235

Phlebotomine sand are delicate, hairy, southeastern states from Louisiana and Arkansas to mosquito-like insects occurring mainly in the tropics Maryland and Delaware. However, collection and subtropics. They are notorious vectors of agents records in this country are spotty, and the species has of several deadly and disfiguring diseases such as been reported in only 45 counties. In Mississippi, it , sand fly fever, and bartonellosis previously was known from only two counties. (Lane 1996). Lutzomyia shannoni (Dyar) is one of Young and Perkins (1984) reported the collection of the more thoroughly studied species of the phleboto- two males and 37 females by V. Newhouse in Han- mine sand flies in North America (Figure 1). The cock County, and Rozeboom (1944) identified an various life stages were figured by Hanson (1968) unknown number of Lu. shannoni from Clinton, and Young and Perkins (1984). Brinson et al. Hinds County, in the collection of the U.S. National (1992) and Comer et al. (1994b) documented the Museum. A search for phlebotomines revealed no seasonal abundance of adults in Georgia. The specimens in either the Mississippi Entomological former authors also documented the vertical distri- Museum at Mississippi State University (R. Brown, bution of adults, while the latter determined there pers. comm.) or the insect collection at the University were three generations per year. Hosts for the of Mississippi (P. K. Lago, pers. comm.). In this blood-feeding females include white-tailed deer, note, we report additional collection records for Lu. feral swine, donkeys, , (Comer et al. shannoni in Mississippi, including three new county 1994a), and humans (Thurman et al. 1949, Snow records (dates preceded by an asterisk). 1955). Ecologically, Lu. shannoni is associated with Centers for Disease Control miniature light traps live oak (Quercus viginiana Miller) forests and, to a supplemented with dry ice were used for all collec- lesser extent, mixed hardwood forests where an tions. Traps were set in areas of mature, mixed oak- abundance of tree holes provides diurnal resting sites for adults (Comer et al. 1993). In the eastern United States, Lu. shannoni is a vector of vesicular stomatitis virus (Corn et al. 1990, Comer et al. 1994b) and, because females transmit the virus transovarially to a percentage of their progeny (Comer et al. 1990), the species provides an overwintering mechanism for the virus. Lawyer and Young (1987) determined Lu. shannoni also is a competent vector of mexicana (Biagi), a parasitic protozoan that may be enzootic in the southeastern United States (McHugh et al. 2003), and unidentified flagellates were detected in Lu. shannoni collected in Florida (Perkins 1982). Lutzomyia shannoni has a discontinuous range Figure 1. Adult sand fly feeding (Photo from Argentina to the United States with a gap in courtesy the U.S. Armed Forces Pest Texas and northern Mexico, possibly due to the Management Board, photo by Dr. Edgar absence of extensive hardwood forests (Young and Rowton). Duncan 1994). In the United States, it occurs in the

1Author for correspondence. Mississippi Department of Health, P.O. Box 1700, Jackson, MS 39215. Telephone 601-576-7689.

July 2005 Vol 50, No. 3 195 hickory-pine forests containing predominantly hard- (Diptera: Psychodidae) in relation to the epizootiology of woods such as red oak, chestnut oak, white oak, vesicular stomatitis virus on Ossabaw Island, Georgia. J. Med. Entomol. 31:850–854. sweet gum, sycamore, various hickory species, and Comer, J.A., D.M. Kavanaugh, D.E. Stallknecht, G.O. Ware, loblolly pine. A large number of the hardwoods J.L. Corn, and V.F. Nettles. 1993. Effect of forest type on contained tree holes. Specimens were identified the distribution of Lutzomyia shannoni (Diptera: Psychodi- using the key of Young and Perkins (1984), and dae) and vesicular stomatitis virus on Ossabaw Island, Georgia. J. Med. Entomol. 30:555–560. voucher specimens were deposited with the Missis- Comer, J.A., R.B. Tesh, G.B. Modi, J.L. Corn, and V.F. sippi Entomological Museum at Mississippi State Nettles. 1990. Vesicular stomatitis virus, New Jersey University, Starkville, Mississippi. serotype: replication in and transmission by Lutzomyia shannoni (Diptera: Psychodidae). Am. J. Trop. Med. Hyg. County records: COPIAH: *2-V-2001, Copiah 42:483–490. Corn, J.L., J.A. Comer, G.A. Erickson, and V.F. Nettles. 1990. County Game Management Area (GMA), J. God- Isolation of vesicular stomatitis virus New Jersey serotype dard, 2 females. 15-XI-2001, Copiah GMA, J. from phlebotomine sand flies in Georgia. Am. J. Trop. Goddard, 2 females. 21-IX-2004, Copiah GMA, J. Med. Hyg. 42:476–482. Goddard, 1 female. 30-IX-2004, Copiah GMA, W. Hanson, W.J. 1968. The immature stages of the subfamily in Panama. Ph.D. Dissertation, University Varnado, 1 female, 1 male. 1-X-2004, Copiah of Kansas, Manhattan. 104 pp. GMA, W. Varnado, 1 female, 2 males. HINDS: 15- Lane, R.P. 1996. Sandflies. In: Medical Insects and Arachnids. VIII-2003, Byram, J. Goddard, 1 female. 6-V-2004, Lane, R.P. and Crosskey, R.W., eds. Chapman and Hall, Byram, J. Goddard, 1 male. LEAKE: *16-IX-2004, London. near Carthage, J. Goddard, 1 female. TISHO- Lawyer, P.G., and D.G. Young. 1987. Experimental transmis- sion of to hamsters by bites of MINGO: *1-VI-2004, Tishomingo State Park, J. phlebotomine sand flies (Diptera: Psychodidae) from the Goddard, 1 female. 4-VI-2004, Tishomingo State United States. J. Med. Entomol. 24:458–462. Park, W. Varnado, 19 females, 8 males. McHugh, C.P., M.L. Thies, P.C. Melby, L.D. Yantis, Jr., R.W. Raymond, M.D. Villegas, and S.F. Kerr. 2003. Short report: a disseminated infection of Leishmania mexicana in Although these records more than double the an eastern woodrat, Neotoma floridana, collected in Texas. known collections in Mississippi, our knowledge of Am. J. Trop. Med. Hyg. 69:470–472. the distribution of Lu. shannoni is still incomplete. Perkins, P.V. 1982. The identification and distribution of The paucity of records for Lu. shannoni specifically, phlebotomine sand flies in the United States with notes on and phlebotomines generally, reflects the scarcity of the biology of two species from Florida (Diptera: Psycho- didae). Ph.D. Dissertation, University of Florida, Gaines- workers intentionally targeting sand flies for collec- ville. 195 pp. tion. Thus, at the present time, the known distribu- Rozeboom, L.E. 1944. Phlebotomus limai Fonseca in the United tions of this and other phlebotomine species repre- States (Diptera: Psychodidae). J. Parasitol. 30:274–275. sent more the distribution of collectors than that of Snow, W.E. 1955. Feeding activities of some blood sucking Diptera with reference to vertical distribution in bottomland the sand flies themselves. forest. Ann. Entomol. Soc. Am. 48:512–521. Thurman, D.C., J.A. Mulrennan, and E.B. Thurman. 1949. LITERATURE CITED Occurrence of Phlebotomus (Neophlebotomus) shannoni Dyar in Florida (Diptera: Psychodidae). J. Parasitol. Brinson, F. J., D.V. Hagan, J.A. Comer, and D. A. Strolein. 35:199–200. 1992. Seasonal abundance of Lutzomyia shannoni (Dip- Young, D.G. and M.A. Duncan. 1994. Guide to the identifica- tera: Psychodidae) on Ossabaw Island, Georgia. J. Med. tion and geographic distribution of Lutzomyia sand flies in Entomol. 29:178–182. Mexico, the West Indies, Central and South America Comer, J.A., W.S. Irby, and D.M. Kavanaugh. 1994a. Hosts of (Diptera: Psychodidae). Mem. Amer. Entomol. Inst. No. 54. Lutzomyia shannoni (Diptera: Psychodidae) in relation to Associated Publishers, Gainesville, FL. 881 pp. vesicular stomatitis virus on Ossabaw Island, Georgia, Young, D.G. and P.V. Perkins. 1984. Phlebotomine sand flies U.S.A. Med. Vet. Entomol. 8:325–330. of North America (Diptera: Psychodidae). Mosq. News Comer, J.A., D.M. Kavanaugh, D.E. Stallknecht, and J.L. 44:263–304. Corn. 1994b. Population dynamics of Lutzomyia shannoni

196 Journal of the Mississippi Academy of Sciences President’s Column

As my term as President of the Mississippi ing, and thanks to dedicated editorial leadership, the Academy of Sciences draws to a close, I would like MAS Journal has continued publish quality scholarly to thank the Members of the Academy and the Board works and rivals the journals published by many of Directors for the opportunity to serve the Acad- of the Academies of Science of much larger States. emy. It has, truly, been an honor and a pleasure. I applaud the editorial board for their efforts, and This year has been exceptionally active and, I look forward to continued success. believe, successful. The annual meeting was held at Of the many persons who assisted me throughout a new venue, which met with an overwhelmingly the year, the efforts of Ms. Cynthia Huff in the MAS positive response. The processes of planning and office were invaluable. It has been her guidance and preparing for the annual meeting were the result of assistance throughout the entire year that made the many hours of diligent work by a large number of year successful, and will help guide the transition to persons, as mentioned in my most recent President’s our new President, Dr. Larry McDaniel. I look column. What was remarkable to me was the team forward to another successful year for MAS under effort that occurred to make the meeting run his leadership. Thank you, again, for the opportunity smoothly; I remain amazed at the dedication of our to serve as President of the Mississippi Academy of members. In addition to a successful annual meet- Sciences.—Sarah Lea McGuire

July 2005 Vol 50, No. 3 197